Table of Contents Chapter 1ÑHARDWARE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1-1 1.1 PIN ASSIGNMENT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1-1 1.1.1 1.2 DESCRIPTION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1-3 1.2.1 1.2.2 1.2.3 1.2.4 1.2.5 1.2.6 1.3 Pin-Signal Assignment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1-1 Resetting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1-3 Clocking . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1-3 Power-Down Mode . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1-4 Power and Grounding . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1-5 Memory Interface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1-6 The Codec Interface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1-13 SPECIFICATIONS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1-18 1.3.1 1.3.2 1.3.3 1.3.4 1.3.5 Absolute Maximum Ratings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Electrical Characteristics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Switching Characteristics—Preliminary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Synchronous Timing Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Timing Diagrams . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1-18 1-18 1-20 1-23 1-25 Chapter 2ÑSOFTWARE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-1 2.1 SYSTEM OPERATION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-1 2.1.1 2.1.2 2.1.3 2.1.4 2.1.5 2.1.6 2.1.7 2.2 Microcontroller interface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-5 Memory Interface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-8 codec interface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-9 ALGORITHM FEATURES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-9 2.3.1 2.3.2 2.3.3 ISD 2-1 2-2 2-2 2-3 2-4 2-4 2-5 PERIPHERALS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-5 2.2.1 2.2.2 2.2.3 2.3 The State Machine . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Command Execution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Event Handling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Message Handling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Tone Generation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Initialization and Configuration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Power-Down Mode . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . VCD (Voice Compression and Decompression) . . . . . . . . . . . . . . . . . . . . . . . 2-10 DTMF Detection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-11 Tone and Energy Detection (Call Progress) . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-12 vii ISD-T360SB 2.3.4 2.3.5 Full-Duplex Speakerphone . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-14 Speech Synthesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-16 2.4 VOICEDSP PROCESSOR COMMANDS—QUICK REFERENCE TABLE . . . . . . . . .2-21 2.5 COMMAND DESCRIPTION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .2-24 Chapter 3ÑSCHEMATIC DIAGRAMS . . . . . . . . . . . . . . . . . . . 3-1 3.1 APPLICATION INFORMATION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .3-1 Chapter 4ÑPHYSICAL DIMENSIONS . . . . . . . . . . . . . . . . . . . 4-1 viii Voice Solutions in Silicon™ ISD-T360SB VoiceDSP™ Digital Speech Processor with Master/Slave, Full-Duplex Speakerphone, Multiple Flash —Advance Information— and ARAM/DRAM Support Preliminary Information The VoiceDSP™ product family combines multiple digital signal processing functions on a single chip for cost-effective solutions in telephony, automotive and consumer applications. The VoiceDSP processor offers necessary features to modern telephony products, such as: high-quality, speech record and playback, electrical and acoustic echo cancellation for full-duplex hands-free speakerphone operation. The ISD-T360SB VoiceDSP can be used in various applications: ¥ ¥ Digital telephony with add-on speech processing: Digital Telephone Answering Machines(DTADs); and full-duplex, handsfree speakerphone operation for ISDN, DECT, Digital Spread Spectrum, and analog cordless applications; CT0/1 Base stations. An add-on chip for corded telephones featuring DTAD functions and/or fullduplex, hands-free speakerphone operation. ¥ Stand-alone digital answering machines with full-duplex, hands-free speakerphone operation. ¥ Voice memo recording ¥ Automotive applications employing fullduplex speakerphone operations for hands-free, in-car communications, and for car status and information announcements. Based on ISD’s unique concept which combines 16-bit DSP (Digital Speech Processor) and 16-bit RISC core technology, the ISD-T360SB is a highperforming chip solution for various applications. To facilitate incorporating the VoiceDSP processor, it features system support functions such as an interrupt control unit, codec interface (master, slave), Microwire interface to the system microcontroller, as well as a memory handler for Flash and DRAM memory devices. Design of high-end, price optimized systems are possible with ISD’s VoiceDSP flexible system interfaces (codec, microcontroller, and memory management support). The ISD-T360SB processor operates as a peripheral controlled by the system microcontroller via an enhanced, serial Microwire interface. The system microcontroller typically controls the analog circuits, buttons and display, as well as activates functions through commands. The VoiceDSP executes these commands and returns status information to the Microcontroller. The VoiceDSP software resides in the on-chip ROM. It includes DSP-based algorithms, system support functions, and a software interface to hardware peripherals. December 1998 ISD · 2045 Hamilton Avenue, San Jose, CA 95125 · TEL: 408/369-2400 · FAX: 408/369-2422 · http://www.isd.com ISD-T360SB FEATURES AT A GLANCE DTAD MANAGEMENT ¥ ¥ Highest quality speech recording in PCM format for music on hold or other OGM (Out Going Message) recording and IVS Selectable high-quality speech compression rate of 5.3 Kbit/s, 9.9 Kbits or 16.8 Kbit/s, plus silence compression with each rate CALL AND DEVICE MANAGEMENT ¥ Digital volume control ¥ Least cost routing support (LCR) ¥ Power-down mode ¥ 3V to 5V selectable power supply ¥ 4.096 MHz operation ¥ DTMF generation and detection ¥ Up to 16 minutes recording on a 4-Mbit Flash ¥ Telephone line functions, including busy and dial tone detection ¥ High-quality music compression for music on hold (16.8 Kbits) ¥ Single tone generation ¥ Programmable message tag for message categorization, e.g., Mailboxes, InComing Messages (ICM), OutGoing Messages (OGM) ¥ DTMF detection during message playback PERIPHERAL CONTROL Codec ¥ Message management ¥ Skip forward or backward during message playback ¥ µ-Law, A-Law, and 16-bit linear codec input support ¥ Variable speed playback ¥ Selectable master/slave codec interface ¥ Real-time clock: Day of Week, Hours, Minutes ¥ Supports two in-coming lines in slave mode without speakerphone for DTAD recording ¥ Multi-lingual speech synthesis using International Vocabulary Support (IVS) ¥ Supports up to 16 user selectable speech channels in slave mode ¥ Vocabularies available in: English, Japanese, Mandarin, German, French and Spanish ¥ Supports long-frame and short-frame codecs ¥ Single/double bit clock rate for slave mode ¥ On-chip codec clock generation ¥ Software Automatic Gain Control (AGC) SPEAKERPHONE ¥ Digital full-duplex speakerphone ¥ Acoustic- and line-echo cancellation ¥ Memory ¥ Continuous on-the-fly monitoring of external and internal conditions (acoustic and line) provides high-quality, hands-free, conversation in a changing environment Supports up to four 4-Mbit, four 8-Mbit, or four 16-Mbit Flash devices from Toshiba or Samsung ¥ Supports up to four 16-Mbit ARAM/DRAM memory devices from Toshiba, Samsung, and Samsung-compatible ¥ Minimum microcontroller control intervention (Launch-and-forget) ¥ The number of messages that can be stored is limited only by memory size ¥ Supports: On, Off, Mute, and Hold functions ¥ Direct access to message memory ii Voice Solutions in Silicon™ ISD-T360SB ¥ Message storage contains all data in a concatenated chain of memory blocks. ¥ Memory mapping and product floor test included ¥ Supports external vocabularies, using Flash memory or expansion ROM Microwire ¥ MICROWIRE slave interface to an external microcontroller ¥ Sophisticated command language to optimize system code size INTERNATIONAL VOCABULARY SUPPORT (IVS) Develop a new vocabulary by ISD’s voice prompt development tool, the ISD-IVS360. This vocabulary development tool supports various languages and including their unique grammar structures. ISD-IVS360, PC-Windows95™-based program, synthesizes recorded .wav files into the ISD-T360SB’s various compression rates (including PCM). ISD’s VoiceDSP products store IVS vocabularies on either Flash memory or expansion ROM memories, thus DTAD manufacturers can design a product for multiple countries, featuring various languages. For more details about IVS, refer to the IVS User’s Guide. For manufacturing recorded voice prompt and speech synthesis, the ISD International Vocabulary Support delivers pre-recorded voice prompts in the same high-quality of the user-recorded speech. For complete control over quality and memory management, the IVS features adjustable speech compressions. In addition, several pre-recorded voice prompt sets are available in various languages for further convenience. Available Languages: ¥ English ¥ Japanese ¥ Mandarin ¥ German ¥ French ¥ Spanish ISD iii ISD-T360SB Figure 1-1: ISD-T360SB Block Diagram—Basic Configuration with Four 4Mb/8Mb/16Mb NAND Flash Devices (Samsung/Toshiba) iv Voice Solutions in Silicon™ ISD-T360SB Figure 1-2: ISD-T360SB Block Diagram—Basic Configuration with Four 4Mb Serial Toshiba Flash Devices ISD v ISD-T360SB Figure 1-3: ISD-T360SB Block Diagram—Basic Configuration with Four 16Mb ARAM/DRAM Devices (Samsung) and IVS EPROM vi Voice Solutions in Silicon™ ISD-T360SB 1—HARDWARE Chapter 1ÑHARDWARE 1.1 PIN ASSIGNMENT 1.1.1 PIN-SIGNAL ASSIGNMENT Table 1-1 shows all the pins, and the signals that use them in different configurations. It also shows the type and direction of each signal. The following sections detail the pins of the ISDT360SB processor. Slashes separate the names of signals that share the same pin. VSS VCCA CAS/MMCLK VCC A7 A8 A6 A5 A4 A3 A2 VCC VSS A1 NC A0 NC NC NC NC Figure 1-1: 80-MQFP Package Connection Diagram 80 79 78 77 76 75 74 73 72 71 70 69 68 67 66 65 64 63 62 61 A9 1 60 VSSA A10 2 59 X2/CLKIN RAS/MMDOUT DWE/MMDIN 3 4 58 X1 57 TST NC D0 5 56 NC 6 55 NC D1 7 54 NC D2 VSS D3 8 53 MWRDY ISD-T360SB 52 MWDOUT 80-MQFP 51 NC VSS 9 10 VCCHI VCC 11 12 D4 13 D5 14 47 MWCLK D6 15 46 MWDIN 50 49 Top View 48 MWRQST VCC D7 16 45 CCLK PC0/A11 17 44 CDIN NC 18 43 CFS0 NC NC 19 42 CDOUT 20 41 CFS1 ISD NC RESET PB7 MWCS PB6 PB5 PB4 VSS PB3 PB2 VCC PB1 PB0 PC7 PC6/EMCS/ENV0 PC4/A15 PC5 PC3/A14 PC1/A12 PC2/A13 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 1-1 ISD-T360SB 1—HARDWARE Table 1-1: VoiceDSP Pin Signal Assignment Pin Name Signal Name Type Description A(0:15) A(0:16) Output Address bits 0 through 16 CAS CAS Output DRAM Column Address Strobe CCLK CCLK I/O Codec Master/slave Clock CDIN CDIN Input Data Input from Codec CDOUT CDOUT Output Data Output to Codec CFS0 CFS0 I/O Codec 0 Frame Synchronization CFS1 CFS1 Output Codec 1 Frame Synchronization D(0:7) D(0:7) I/O Data bits 0 through 7 DWE DWE Output DRAM Write Enable EMCS/ENV0 EMCS Output Expansion Memory Chip Select EMCS/ENV0 ENV0 Input Environment Select MMCLK MMCLK Output Master MICROWIRE Clock MMDIN MMDIN Input Master MICROWIRE Data Input MMDOUT MMDOUT Output Master MICROWIRE DATA Output MWCLK MWCLK Input MICROWIRE Clock MWCS MWCS Input MICROWIRE Chip Select MWDIN MWDIN Input MICROWIRE Data Input MWDOUT MWDOUT Output MICROWIRE DATA Output MWRDY MWRDY Output MICROWIRE Ready MWRQST MWRQST Output MICROWIRE Request Signal PB(0:7)3 PB(0:7) I/O Port B, bits 0 through 7 PC(0:7) PB(0:7) I/O Port C, bits 0 through 7 RAS RAS Output DRAM Row Address Strobe RESET RESET Input Reset TST TST Input Test pin VCC VCC Power 3.3 V power supply pin VCCA VCCA Power 3.3 V analog circuitry power supply pin VCCHI VCCHI Power 5 V power supply pin. Connect to VCC if 3.3 V power supply is used. VSS VSS Power Ground for on-chip logic and output drivers VSSA VSSA Power Ground for on-chip analog circuitry X1 X1 Oscillator Crystal Oscillator Interface X2/CLKIN X2 Oscillator Crystal Oscillator Interface 1. 2. 3. 4. 5. TTL1 output signals provide CMOS levels in the steady state, for small loads. Input during reset. CMOS level input. Virtual address lines for IVS ROM. Chip select lines for Serial Flash devices. Schmitt trigger input. 1-2 Voice Solutions in Silicon™ ISD-T360SB 1—HARDWARE 1.2 DESCRIPTION This section provides details of the functional characteristics of the VoiceDSP processor. It is divided into the following sections: If the load on the ENV0 pin causes the current to exceed 10 µA, use an external pull-up resistor to keep the pin at 1. • Resetting Figure 1-2 shows a recommended circuit for generating a reset signal when the power is turned on. • Clocking • Power-Down Mode • Power and Grounding • Memory Interface Figure 1-2: Recommended Power-On Reset Circuit • Codec Interface VCC VCC 1.2.1 ISD-T360 RESETTING The RESET pin is used to reset the VoiceDSP processor. RESET On application of power, RESET must be held low for at least tpwr after VCC is stable. This ensures that all on-chip voltages are completely stable before operation. Whenever RESET is applied, it must also remain active for not less than t RST, see Table 1-10 and Table 1-11. During this period, and for 100 ms after, the TST signal must be high. This can be done with a pull-up resistor on the TST pin. The value of MWRDY is undefined during the reset period, and for 100 ms after. The microcontroller should either wait before polling the signal for the first time, or the signal should be pulled high during this period. Upon reset, the ENV0 signal is sampled to determine the operating environment. During reset, the EMCS/ENV0 pin is used for the ENV0 input signals. An internal pull-up resistor sets ENV0 to 1. After reset, the same pin is used for EMCS. System Load on ENV0 For any load on the ENV0 pin, the voltage should not drop below VENVh (see Table 1-10 and Table 1-11). ISD VSS 1.2.2 CLOCKING The VoiceDSP processor provides an internal oscillator that interacts with an external clock source through the X1 and X2/CLKIN pins. Either an external single-phase clock signal, or a crystal oscillator, may be used as the clock source. External Single-Phase Clock Signal If an external single-phase clock source is used, it should be connected to the CLKIN signal as shown in Figure 1-3, and should conform to the voltage-level requirements for CLKIN stated in “ELECTRICAL CHARACTERISTICS” on page 1-18. 1-3 ISD-T360SB 1—HARDWARE Figure 1-3: External Clock Source Table 1-2 lists the components in the crystal oscillator circuit Table 1-2: VoiceDSP X1 X2/CLKIN Components of Crystal Oscillator Circuit Component Values Tolerance Crystal Resonator 4.096 MHz Resistor R1 10 MΩ 5% Capacitors C1, C2 33 pF 20% Single-phase Clock Signal Clock Generator Crystal Oscillator A crystal oscillator is connected to the on-chip oscillator circuit via the X1 and X2 signals, as shown in Figure 1-4. Figure 1-4: Connections for an External Crystal Oscillator X1 R1 C2 C1 Keep stray capacitance and inductance, in the oscillator circuit, as low as possible. The crystal resonator, and the external components, should be as close to the X1 and X2/CLKIN pins as possible, to keep the trace lengths in the printed circuit to an absolute minimum. You can use crystal oscillators with maximum load capacitance of 20 pF, although the oscillation frequency may differ from the crystal’s specified value. 1-4 POWER-DOWN MODE Power-down mode is useful during a power failure or in a power-saving model, when the power source for the processor is a backup battery or in battery-powered devices, while the processor is in idle mode. In power-down mode, the clock frequency of the VoiceDSP processor is reduced and some of the processor modules are deactivated. As a result, the ISD-T360SB consumes considerably less power than in normal-power mode. Although the VoiceDSP processor does not perform all its usual functions in power-down mode, it does retain stored messages and maintain the date and time. ISD-T360 X2 1.2.3 NOTE In power-down mode all the chip select signals, CS0 to CS3, are set to 1. To guarantee that there is no current flow from these signals to the Flash devices, the power supply to these devices must not be disconnected. The ISD-T360SB stores messages and all memory management information in Flash or ARAM/ DRAM memory. When Flash memory is used for memory management, power does not need to be maintained to the processor to preserve stored messages. When ARAM/DRAM memory is used for message management, preserving stored messages requires a battery back up dur- Voice Solutions in Silicon™ ISD-T360SB 1—HARDWARE ing a power failure. If the power failure extends the life of the battery, the microcontroller should perform the initialization sequence (as described on page 2-4), and use the SETD command to set the date and time. Figure 1-5 when operating in a 3.3 V environment, and as shown in Figure 1-6 when operating in a 5 V environment. Failure to correctly connect the pins may result in damage to the device. To keep power consumption low during powerdown mode, the RESET, MWCS, MWCLK and MWDIN signals should be held above V CC – 0.5 V or below VSS + 0.5 V. The Capacitor and Resistor values are given in Table 1-3. Table 1-3: Components of Supply Circuit Component 1.2.4 POWER AND GROUNDING Power Pin Connections The ISD-T360 can operate over two supply voltage ranges 3.3 V ±10% and 5 V ±10%. The five power supply pins (VCC, VSS, VCCA, VSSA and VCCHI) must be connected as shown in Values Tolerance Resistor R1, R2 10 Ω 5% Capacitors(1) C1, C2 Capacitors C3, C4, C5, C6, C7 1.0 µF Tantalum 0.1 µF Ceramic 20% 1. All capacitors represent two parallel capacitors at the values 1.0 µF and 0.1 µF. Figure 1-5: 3.3 V Power Connection Diagram R1 3.3 V Supply VSS VSS VCCHI VCC VCC VCC VCCA VSS 74 72 11 63 ISD-T360 12 61 VSSA 60 50 48 30 VCC ISD 64 9 C1 VSS VCC 32 VSS 1-5 ISD-T360SB 1—HARDWARE Figure 1-6: 5 V Power Connection Diagram C6 R2 C5 VSS VSS VCC VCC VCCA 74 72 64 VSS C7 63 9 61 V A 60 SS 5 V Supply C2 VCCHI 11 VCC 12 50 ISD-T360 VSS C4 48 30 32 VCC VCC VSS C3 For optimal noise immunity, the power and ground pins should be connected to V CC and the ground planes, respectively, on the printed circuit board. If VCC and the ground planes are not used, single conductors should be run directly from each VCC pin to a power point, and from each GND pin to a ground point. Avoid daisychained connections. The VoiceDSP does not perform its usual functions in power-down mode but it still preserves stored messages, maintains the time of day and generates ARAM/DRAM refresh cycles. When you build a prototype, using wire-wrap or other methods, solder the capacitors directly to the power pins of the VoiceDSP processor socket, or as close as possible, with very short leads. 1-6 1.2.5 MEMORY INTERFACE Flash Support The ISD-T360SB VoiceDSP supports Flash devices for storing recorded data, thus, power can be disconnected to the ISD-T360SB without losing data. The ISD-T360SB supports serial and semiparallel Flash device interfaces, such as TC58V16BFT, TC5816BFT, TC58A040F, KM29N040T, KM2928000T/IT, and KM29216000AT/AIT. The ISDT360SB may be connected to up to four Flash devices, resulting with maximum recording storage of 16-Mbits x 4 = 64 Mbits (up to 4 hours of recording time). The following flash devices are supported: Voice Solutions in Silicon™ ISD-T360SB 1—HARDWARE Table 1-4: Supported Flash Devices Manufacturer Memory Device Name Characteristics Operating Voltage Memory Size Conversion Type Toshiba TC58V16BFT 2Mx8 3.3 V 16-Mbit LV µ-Law Toshiba TC5816BFT 2Mx8 5V 16-Mbit A-Law Toshiba TC58A040F Serial 5V 4-Mbit µ-Law, A-Law Samsung KM29N040T 512Kx8 5V 4-Mbit µ-Law, A-Law, LV Samsung KM29W8000T/IT 1Mx8 5V 8-Mbit µLaw Samsung KM29W16000AT /AIT 2Mx8 5V 16-Mbit A-Law Internal Memory Organization Toshiba Serial Flash The Flash devices detailed in Table 1-4 divide internally into basic 4-Kbyte block units. The ISDT360SB uses one block on each device for memory management, leaving the rest of the blocks available for recording. Using at least one block for a single recorded message yields a maximum of NUM_OF_BLOCKS_IN_MEM – 1 (see Table 2-10 for parameter definition) messages per device. The VoiceDSP processor supports up to four TC58A040F, 4Mbit, serial interface, Flash memory devices for storing messages. The TC58A040F is organized as an array of 128 blocks, with a dedicated, read-only, bad block map programmed by the manufacturer and located in the last block. This map is used by the ISD-T360SB to define the available blocks for recording. The ISDT360SB uses the VoiceDSP master MICROWIRE interface to communicate, serially, with the Flash devices, while selecting the current Flash device using PB3-PB6. Connecting less than four Flash devices require connecting the Flash devices sequentially, starting from PB3 up to PB6 (see Figure 1-7). Refer to Figure 1-34for the Master MICROWIRE timing diagram. Upon initialization the ISD-T360SB activates a sifting algorithm to detect defected blocks. Defected blocks are defined as blocks with over 10 bad nibbles (a nibble consists of 4 bits). The defected blocks are marked as UNUSED and excluded from the list of available blocks for recording. ISD 1-7 ISD-T360SB 1—HARDWARE Figure 1-7: Memory Interface with Four Toshiba 4Mbit Serial Flash Devices and Optional Voice IVS EPROM 1-8 Voice Solutions in Silicon™ 1—HARDWARE ISD-T360SB Figure 1-8: Memory Interface with Four 4MB/16 Mbit, NAND Flash Devices (Samsung, Toshiba) NAND Flash (Samsung, Toshiba) The VoiceDSP processor supports up to four, semi-parallel interface, Flash memory devices for storing messages. Flash device with semi-parallel interface uses a single 8bit I/O port to set the address and access the data. The ISD-T360SB supports three types of Flash volumes (4Mbit, 8Mbit and 16Mbit as listed in Table 1-4) while all the connected Flash devices must be of the same type. Ports B and C are used to connect ISDT360SB to the Flash devices using port B for address and data transfer and port C for communication control and chip select. Connecting less than four Flash devices require connecting the Flash devices sequentially, starting from PC4 up to PC7 (see Figure 1-8). The ISD-T360SB scans the Flash devices upon initialization, sifting out the bad blocks, and marking them in a special map, located in the last block of each device. ISD 1-9 ISD-T360SB 1—HARDWARE Figure 1-9: Memory Interface with Four Toshiba 4Mbit Serial Flash Devices and Optional IVS EPROM Flash Endurance A Flash memory may be erased a limited number of times. To maximize the Flash use, the memory manager utilizes the Flash’s blocks evenly (i.e., each block is erased more or less the same number of times), to ensure that all blocks have the same lifetime. Refer to the respective Flash memory device data sheets for specific endurance specifications. A VoiceDSP processor message uses at least one block. 1-10 The maximum recording time depends on four factors: 1. The basic compression rate (5.3 Kbit/s, 9.9Kbit/s, or 16.8Kbit/s). 2. The amount of silence in the recorded speech. 3. The number of bad blocks. Voice Solutions in Silicon™ ISD-T360SB 1—HARDWARE The number of recorded messages. (The basic memory allocation unit for a message is a 4Kbyte block, which means that half a block on average is wasted per recorded message). Table 1-5: Recording Time with 15% Silence Compression Compression Rate Total Recording Time 4 Mbit 5.3 Kbit/s 14.9 Minutes 4 Mbit 9.9 Kbit/s 8.1 Minutes 4 Mbit 16.8 Kbit/s 4.8 Minutes 8 Mbit 5.3 Kbit/s 29.8 Minutes 8 Mbit 9.9 Kbit/s 16.2 Minutes 8 Mbit 16.8 Kbit/s 9.6 Minutes 16 Mbit 5.3 Kbit/s 59.6 Minutes 16 Mbit 9.9 Kbit/s 32.4 Minutes 16 Mbit 16.8 Kbit/s 19.2 Minutes 32 Mbit 5.3 Kbit/s 119.2 Minutes 32 Mbit 9.9 Kbit/s 64.8 Minutes 32 Mbit 16.8 Kbit/s 38.4 Minutes Memory Size ARAM/DRAM Support The VoiceDSP processor supports up to four, 16-Mbit, ARAM/DRAM devices for storing messages. The ISD-T360 connects to the ARAM/ DRAM device using address buses A0–A11 and data buses D0–D3. This connection allows access to 222 nibbles (16-Mbit) on each device. The ISD-T360SB selects the current ARAM/DRAM device using PB3–PB6 as described in Figure 1-10. Using less than four ARAM/DRAM devices requires connecting the devices sequentially, starting from PB3 up to PC6. RAS and CAS are connected in parallel to all the ARAM/DRAM devices and are used to refresh the memory. The difference between ARAM and DRAM resides with the amount of bad cells on the device and the device performance. While DRAM has no bad cells, ARAM contains certain level of impurity and thus requires testing and mapping of the bad blocks upon the initialization of the ISDT360SB. Although there are no real blocks on the ISD ARAM device, the ISD-T360SB emulates virtual “blocks” on the ARAM device (as if it was a Flash device), tests these “blocks” and marks them in a special map on the last “block” on each device. This test is required only when using ARAM devices (as opposed to DRAM devices that require no testing due to lack of bad blocks). The virtual division to blocks simplifies the use of ARAM/DRAM devices and allows the use of the same set of commands for Flash and ARAM/ DRAM. Another major difference between ARAM/ DRAM and Flash devices is that the internal mapping in the ARAM is lost upon power off. Thus, the initialization process needs to take place after each power reset. The mapping is not lost when entering and exiting the Power-Down Mode. Refer to Figure 1-24 through Figure 1-28 for Timing Diagrams of ARAM/DRAM Read, Write, Refresh in Normal Mode and Refresh in Power-Down Mode Cycles. ROM Interface IVS vocabularies can be stored in either Flash memory and/or ROM. The VoiceDSP processor supports IVS ROM devices through an Expansion Memory mechanism. Up to 64 Kbytes (64K x 8) of Expansion Memory are directly supported. Nevertheless, the processor uses bits of the on-chip port (PB) to further extend the 64 Kbytes address space up to 0.5 Mbytes address space. ROM is connected to the VoiceDSP processor using the data bus, D(0:7), the address bus, A(0:15), the extended address signals, EA(16:18), and Expansion Memory Chip Select, EMCS, controls. The number of extended address pins to use may vary, depending on the size and configuration of the ROM. ISD-T360SB configured with semi-parallel Flash memory can not support extension ROM. 1-11 ISD-T360SB 1—HARDWARE Figure 1-10: Memory Interface with Four 16-Mbit ARAM/DRAM Devices (Samsung, Toshiba) and Optional IVS EPROM Table 1-6: Supported DRAM Devices Manufacturer Memory Device Name Characteristics Operating Voltage Memory Size Samsung KM44C4004CS-6 EDO 4Mx4 5V 16-Mbit EDO LV Samsung KM44V4004CS-6 EDO 4Mx4 3.3 V 16-Mbit EDO Toshiba TP EDO 4Mx4 5V, 3.3V 16-Mbit EDO, LV 1-12 Voice Solutions in Silicon™ ISD-T360SB 1—HARDWARE 1.2.6 THE CODEC INTERFACE The ISD-T360 provides an on chip interface for analog and digital telephony, supporting master and slave codec interface modes. In master mode, the ISD-T360 controls the operation of the codec for use in analog telephony. In the slave mode, the ISD-T360 codec interface is controlled by an external source. This mode is used in digital telephony (i.e., ISDN or DECT lines). The slave mode is implemented with respect to IOM-2™/ CGI specifications. See Table 1-7 for codec options for the ISDT360SB (ISD supports compatible codecs in addition to those listed below). The codec interface supports the following features: • Master Mode or Slave Mode. • 8- or 16-bit Channel Width. • Long (Variable) or Short (Fixed) Frame Protocol. • Single or Double Bit (Slave Mode Only) Clock Rate. CDIN, CDOUT, CCLK, and CFS1 pins. Data is transferred to the codec through the CDOUT output pin. Data is read from the codec through the CDIN input pin. The CCLK and CFS0 pins are output in Master Mode and input in Slave Mode. The CFS1 is an output pin. Short Frame Protocol When the short frame protocol is configured, eight or sixteen data bits are exchanged with each codec in each frame (i.e., the CFS0 cycle). Data transfer begins when CFS0 is set to 1 for one CCLK cycle. The data is then transmitted, bit by bit, via the CDOUT pin. Concurrently, the received data is shifted in through the CDIN pin. Data is shifted one bit per CCLK cycle. After the last bit has been shifted, CFS1 is set to 1 for one CCLK cycle. Then, the data from the second codec is shifted out via CDOUT, concurrently with the inward shift of the data received via CDIN. Long Frame Protocol • Single or Dual Channel Codecs • One or Two Codecs • Multiple Clock And Sample Rates. • One or Two Frame Sync Signals This codec interface uses five signals: CDIN, CDOUT, CCLK, CFS0, and CFS1. The CDIN, CDOUT, CCLK, and CFS0 pins are connected to the first codec. The second codec is connected to When long frame protocol is configured, eight or sixteen data bits are exchanged with each codec, as for the short frame protocol. However, for the long frame protocol, data transfer starts by setting CFS0 to 1 for eight or sixteen CCLK cycles. Short or long frame protocol is available in both Master and Slave modes. Figure 1-11 illustrates an example of short frame protocol with an 8-bit channel width. Table 1-7: Supported Codec Devices Manufacturer Codec Device Name Characteristics Operating Voltage Conversion Type National Semiconductor TP3054 Single codec 5V µ-Law National Semiconductor TP 3057 Single codec 5V A-Law Oki MSM7533V Dual codec 5V µ-Law, A-Law Oki MSM 7704 Dual codec 3.3 V µ-Law, A-Law, LV Macronix MX93002FC Dual codec 5V µ-Law Lucent T7502 Dual codec 5V A-Law Lucent T7503 Dual codec 5V µ-Law ISD 1-13 ISD-T360SB 1—HARDWARE Figure 1-11: Codec Protocol-Short Frame—8-Bit Channel Width Channel Width The Codec interface supports both 8-bit and 16bit channel width in Master and Slave Modes. Slave Mode The VoiceDSP supports digital telephony applications including DECT and ISDN by providing a Slave Mode of operation. In Slave Mode operation, the CCLK signal is input to the ISD-T360 and controls the frequency of the codec interface operation. The CCLK may take on any frequency between 500 KHz and 4 MHz. Both long and short frame protocol are supported with only the CFS1 output signal width affected. The CFS0 input signal must be a minimum of one CCLK cycle. This interface supports ISDN protocol with one bit clock rate or double bit clock rate. The exact format is selected with the CFG command. The slave codec interface uses four signals: CDIN, CDOUT, CCLK, and CFS0. The CDIN, CCLK, and CFS0 input pins and the CDOUT output pins are connected to the ISDN/DECT agent. Data is transferred to the VoiceDSP through the CDIN pin and read out through the CDOUT pin. The CFS0 pin is used to define the start of each frame (see below) the source of that signal is at the master side. The CCLK is used for bit timing of CDIN and CDOUT. The rate of the CCLK is configured via the CFG command and can be twice of the data rate or at the data rate. The source of that signal is at the master side. In slave mode, a double clock bit rate feature is available as well. When the codec interface is configured to double clock bit rate, the CCLK input signal is divided internally by two and the resulting clock used to control the frequency of the codec of the codec interface operation. 1-14 Voice Solutions in Silicon™ ISD-T360SB 1—HARDWARE Table 1-8: Typical Codec Applications Application Codec Type No. of Channels Master/ Slave Channel Width (No. Bits) Long/ CCLK Short Bit Rate Freq. Frame (MHz) Protocol Sample Rate (Hz) No. of Frame Syncs Analog µLaw single 1 Master 8 short or long 1 2.048 8000 1 ISDN—8 bit digital—ALaw dual 2 Slave 8 short 1 or 2 2.048 8000 1 Linear single 1 Master 16 short 1 2.048 8000 1 IOM-2/GCI single or dual 1–2 Slave 8 short 1 or 2 1.536 8000 1 266 Compatibility single or dual 1 or 2 Master 8 long or short 1 2.048 8000 1 or 2 Figure 1-12: Codec Interface with One Single Codec, NSC TP3054, for Single Line Operation ISD 1-15 ISD-T360SB 1—HARDWARE Figure 1-13: Codec Interface Diagram with Two, Single Codecs, NSC TP3054, and NSC TP3057, for Speakerphone Operation Figure 1-14: Codec Interface for Dual Line or Single Line and Speakerphone Operation with OKI Dual Codec 1-16 Voice Solutions in Silicon™ 1—HARDWARE ISD-T360SB Figure 1-15: Codec Interface for Dual Line or Single Line and Speakerphone with Lucent Dual Codec Figure 1-16: Codec Interface for Dual Line or Single Line and Speakerphone Operation with Macronix Dual Codec ISD 1-17 ISD-T360SB 1—HARDWARE 1.3 SPECIFICATIONS 1.3.1 ABSOLUTE MAXIMUM RATINGS Storage temperature –65˚C to +150˚C Temperature under bias 0˚C to +70˚C All input and output voltages with respect to GND –0.5 V to +6.5 V 1.3.2 NOTE Absolute maximum ratings indicate limits beyond which permanent damage may occur. Continuous operation at these limits is not intended; operation should be limited to the conditions specified below. ELECTRICAL CHARACTERISTICS TA = 0ºC to +70ºC, VCC = 5 V ±10% Or 3.3 V ±10%, GND = 0 V Table 1-9: Electrical Characteristics—Preliminary Information (All Parameters with Reference to VCC = 3.3 V) Symbol Parameter Conditions Min Capacitance1 Typ CX X1 and X2 ICC1 Active Supply Current Normal Operation Mode, Running Speech Applications2 40.0 ICC2 Standby supply current Normal Operation Mode, DSPM Idle2 30.0 ICC3 Power-down Mode Supply Current Power-down Mode2,3 IL Input Load Current 0 V ≤ VIN ≤ VCC IO (Off) Output Leakage Current (I/O pins in Input Mode) 0 V ≤ VOUT ≤ VCC tCASa CAS Active After R.E. CTTL, T1 or T2W3 tCASh CAS Hold After R.E. CTTL tCASia CAS Inactive After R.E. CTTL, T3 or TERF tCASLw DRAM, PDM, CAS Width At 0.8 V, Both Edges Max 17.0 pF 80.0 mA mA 0.7 mA –5.0 5.0 µA –5.0 5.0 µA 12.0 0.0 12.0 600.0 tDWEa DWE Active After R.E. CTTL, T2W2 12.0 tDWEh DWE Hold After R.E. CTTL 12.0 tDWEia DWE Inactive After R.E. CTTL, T3 12.0 tRASa RAS Active After R.E. CTTL, T2W1 or T2WRF 12.0 tRASh RAS Hold After R.E. CTTL tRASia RAS Inactive After R.E. CTTL, T3 or T3RF tRASLw DRAM PDM, RAS Width At 0.8 V, Both Edges 200.0 tRLCL DRAM PDM RAS Low, after CAS Low F.E. CAS to F.E. RAS 200.0 tWRa WR0 Active After R.E. CTTL, T1 1-18 Units 0.0 12.0 tCTp/2+2 Voice Solutions in Silicon™ ISD-T360SB 1—HARDWARE Table 1-9: Electrical Characteristics—Preliminary Information (All Parameters with Reference to VCC = 3.3 V) Symbol Parameter Conditions 4 Min Typ Max Units tWRCSh WR0 Hold after EMCS R.E. EMCS R.E. to R.E. WR0 tWRh WR0 Hold After R.E. CTTL tWRia WR0 Inactive After R.E. CTTL, T3 VENVh ENV0 Input, High Voltage 2.0 V VHh CMOS Input with Hysteresis, Logical 1 Input Voltage 2.1 V VHl CMOS Input with Hysteresis, Logical 0 Input Voltage VHys Hysteresis Loop Width1 0.5 VIH TTL Input, Logical 1 Input Voltage 2.0 VCC+0.5 V VIL TTL Input, Logical 0 Input Voltage –0.5 0.8 V VOH Logical 1 TTL, Output Voltage IOH = –0.4 mA 2.4 V VOHWC MMCLK, MMDOUT and EMCS Logical 1, Output Voltage IOH = –0.4 mA 2.4 V µA5 VCC–0.2 V VOL VOLWC Logical 0, TTL Output Voltage IOH = –50 IOL = 4 mA IOL = 4 mA VXH CLKIN Input, High Voltage External Clock VXL CLKIN Input, Low Voltage External Clock 5. ISD tCTp/2+2 IOL = 50 µA5 IOL = 50 µA5 3. 4. tCTp/2–6 0.8 MMCLK, MMDOUT and EMCS Logical 0, Output Voltage 1. 2. 10.0 V V 0.45 V 0.2 V 0.45 V 0.2 V 2.0 V 0.8 V Guaranteed by design. IOUT =0, TA 25˚C, VCC = 3.3 V for VCC pins and 3.3 V or 5 V on VCCHI pins, operating from a 4.096 MHz crystal and running from internal memory with Expansion Memory disabled. All input signals are tied to 0 (above VCC – 0.5 V or below VSS + 0.5 V), except ENV0, which is tied to VCC. Measured in power-down mode. The total current driven, or sourced, by all the VoiceDSP processor’s output signals is less than 50 µA. Guaranteed by design, but not fully tested. 1-19 ISD-T360SB 1.3.3 1—HARDWARE SWITCHING CHARACTERISTICS—PRELIMINARY Definitions All timing specifications in this section refer to 0.8 V or 2.0 V on the rising or falling edges of the signals, as illustrated in Figure 1-17 through Figure 1-23, unless specifically stated otherwise. Maximum times assume capacitive loading of 50pF. CLKIN crystal frequency is 4.096 MHz. NOTE CTTL is an internal signal and is used as a reference to explain the timing of other signals. See Figure 1-37. Figure 1-17: Synchronous Output Signals (Valid, Active and Inactive) 2.0V CTTL or MWCLK 2.0V Signal 0.8V tSignal NOTE: Signal valid, active or inactive time, after a rising edge of CTTL or MWCLK. Figure 1-18: Synchronous Output Signals (Valid) MWCLK 0.8V 2.0V Signal 0.8V tSignal NOTE: 1-20 Signal valid time, after a falling edge of MWCLK. Voice Solutions in Silicon™ ISD-T360SB 1—HARDWARE Figure 1-19: Synchronous Output Signals (Hold), after Rising Edge of CTTL 2.0V CTTL 2.0V Signal 0.8V tSignal NOTE: Signal hold time, after a rising edge of CTTL. Figure 1-20: Synchronous Output Signals (Hold), after Falling Edge of MWCLK 2.0V MWCLK 2.0V Signal 0.8V tSignal NOTE: Signal hold time, after a falling edge of MWCLK. Figure 1-21: Synchronous Input Signals CTTL or MWCLK 2.0 V 2.0 V 2.0 V 0.8 V 0.8 V Signal tSignal NOTE: ISD Setup tSignal Hold Signal setup time, before a rising edge of CTTL or MWCK, and signal hold time after a rising edge of CTTL or MWCK 1-21 ISD-T360SB 1—HARDWARE Figure 1-22: Asynchronous Signals 2.0 V Signal A 0.8 V 2.0 V Signal B 0.8 V tSignal NOTE: Signal B starts after rising or falling edge of signal A. The RESET has a Schmitt trigger input buffer. Figure 1-23 shows the input buffer characteristics. Figure 1-23: Hysteresis Input Characteristics Vout VHys VHl 1-22 VHh Voice Solutions in Silicon™ ISD-T360SB 1—HARDWARE 1.3.4 SYNCHRONOUS TIMING TABLES In this section, R.E. means Rising Edge and F.E. means Falling Edge. Table 1-10: Output Signals—Preliminary Symbol Figure Description Reference Conditions Min (ns) Max (ns) tAh Address Hold After R.E. CTTL tAv Address Valid After R.E. CTTL, T1 12.0 tCCLKa CCLK Active After R.E. CTTL 12.0 tCCLKh CCLK Hold After R.E. CTTL tCCLKia CCLK Inactive After R.E. CTTL tCDOh CDOUT Hold After R.E. CTTL tCDOv CDOUT Valid After R.E. CTTL Period1 0.0 12.0 0.0 12.0 tCTp CTTL Clock tEMCSa EMCS Active After R.E. CTTL, T2W1 tEMCSh EMCS Hold After R.E. CTTL tEMCSia EMCS Inactive After R.E. CTTL T3 12.0 tFSa CFS0 Active After R.E. CTTL 25.0 tFSh CFS0 Hold After R.E. CTTL tFSia CFS0 Inactive After R.E. CTTL tMMCLKa Master MICROWIRE Clock Active After R.E. CTTL tMMCLKh Master MICROWIRE Clock Hold After R.E. CTTL tMMCLKia Master MICROWIRE Clock Inactive After R.E. CTTL tMMDOh Master MICROWIRE Data Out Hold After R.E. CTTL tMMDOv Master MICROWIRE Data Out Valid After R.E. CTTL 12.0 tMWDOf MICROWIRE Data Float1 After R.E. MWCS 70.0 tMWDOh tMWDOnf R.E. CTTL to next R.E. CTTL 0.0 MICROWIRE Data Out MICROWIRE Data No Hold2 Float2 Out Valid2 30.5 250,000 12.0 0.0 0.0 25.0 12.0 0.0 12.0 0.0 After F.E. MWCLK 0.0 After F.E. MWCS 0.0 70.0 tMWDOv MICROWIRE Data tMWITOp MWDIN to MWDOUT tMWRDYa MWRDY Active After R.E. of CTTL 0.0 35.0 tMWRDYia MWRDY Inactive After F.E. MWCLK 0.0 70.0 tPABCh PB and MWRQST After R.E. CTTL 0.0 tPABCv PB and MWRQST After R.E. CTTL, T2W1 1. 2. ISD After F.E. MWCLK 70.0 Propagation Time 70.0 12.0 In normal operation mode, tCTp must be 48.8 ns; in power-down mode, tCTp must be 50,000 ns. Guaranteed by design, but not fully tested. 1-23 ISD-T360SB 1—HARDWARE Table 1-11: Input Signals—Preliminary Symbol Figure Description Reference Conditions Min (ns) tCCLKSp Codec Clock Period (slave) R.E. CCLK to next R.E. CCLK 244 tCCLKSh Codec Clock High (slave) At 2.0 V (both edges) 120 tCCLKSl Codec Clock Low (slave) At 0.8 V (both edges) 120 tCDIh CDIN Hold After R.E. CTTL 0.0 tCDIs CDIN Setup Before R.E. CTTL 11.0 tCFS0Ss CFS0 Setup Before R.E. CCLK TBD tCFS0Sh CFS0 Hold After R.E. CCLK TBD tDIh Data in Hold (D0:7) After R.E. CTTL T1, T3 or TI 0.0 tDIs Data in Setup (D0:7) Before R.E. CTTL T1, T3 or TI 15.0 tMMDINh Master MICROWIRE Data In Hold After R.E. CTTL 0.0 tMMDINs Master MICROWIRE Data In Setup Before R.E. CTTL 11.0 tMWCKh MICROWIRE Clock High (slave) At 2.0 V (both edges) 100.0 tMWCKl MICROWIRE Clock Low (slave) At 0.8 V (both edges) 100.0 tMWCKp MICROWIRE Clock Period (slave)1 R.E. MWCLK to next R.E. MWCLK 2.5 µs tMWCLKh MWCLK Hold After MWCS becomes inactive 50.0 tMWCLKs MWCLK Setup Before MWCS becomes active 100.0 tMWCSh MWCS Hold After F.E. MWCLK 50.0 tMWCSs MWCS Setup Before R.E. MWCLK 100.0 tMWDIh MWDIN Hold After R.E. MWCLK 50.0 tMWDIs MWDIN Setup Before R.E. MWCLK 100.0 tPWR Power Stable to RESET R.E.2 After VCC reaches 4.5 V 30.0 ms tRSTw RESET Pulse Width At 0.8 V (both edges) 10.0 ms tXh CLKIN High At 2.0 V (both edges) tX1p/2 – 5 tXl CLKIN Low At 0.8 V (both edges) tX1p/2 – 5 tXp CLKIN Clock Period R.E. CLKIN to next R.E. CLKIN 1. 2. Max (ns) 244.4 Guaranteed by design, but not fully tested in power-down mode. Guaranteed by design, but not fully tested. 1-24 Voice Solutions in Silicon™ ISD-T360SB 1—HARDWARE 1.3.5 TIMING DIAGRAMS Figure 1-24: ROM Read Cycle Timing 1. 2. 3. This cycle may be either TI (Idle), T3 or T3H. Data can be driven by an external device at T2W1, T2W, T2 and T3. This cycle may be either TI (Idle) or T1. Figure 1-25: ARAM/DRAM Refresh Cycle Timing (Normal Operation) 1. ISD This cycle may be either TI (Idle) or T1 of any non-DRAM bus cycle. If the next bus cycle is a DRAM one, T3RF is followed by three TI (Idle) cycles. 1-25 ISD-T360SB 1—HARDWARE Figure 1-26: ARAM/DRAM Power-Down Refresh Cycle Timing Figure 1-27: DRAM Read Cycle Timing TI T1 T2W1 T2W2 T2W3 6xT2W T2 T3 TI TI (Note 2) TI T1 T2W1 T2W2 T2W3 CTTL RAS tRASa tRASia tRASh tRASh CAS A0-10 or A0-15 (Note 1) tAv Row tCASa tCASia tCASh tCASh Column tAv tAh tAh DWE (Note 3) D0-1, D3-7 Data In tDf tDIs tDIh (Note 3) D2/RA11 1. 2. 3. RA11 D2 In RA11 tDf tDf tDIs tDIh tAv A0-A10 in the IRE environment, when Expansion Memory is disabled; otherwise A0-A15. This cycle may be either TI (Idle) or T1 of any non-DRAM bus cycle. If the next bus cycle is to DRAM, T3 is followed by three TI (Idle) cycles. An external device can drive data from T2W3 to T3. 1-26 Voice Solutions in Silicon™ ISD-T360SB 1—HARDWARE Figure 1-28: DRAM Write Cycle Timing TI T1 T2W1 T2W2 T2W3 6xT2W T2 T3 TI TI (Note 2) CTTL RAS tRASa tRASia tRASh tRASh CAS A0-10 or A0-15 (Note 1) tCASa tCASia tCASh tCASh Row Column tAv tAh tAv DWE tDWEa tDWEia tDWEh tDWEh D0-1, D3-7 Data Out tDf tDh tDv tDf RA11 D2/RA11 tAv D2 Out tDv tDh tDf 1. 2. ISD A0–A10 in the IRE environment, when Expansion Memory is disabled; otherwise A0–A15. This cycle may be either TI (Idle) or T1 of any non-DRAM bus cycle. If the next bus cycle is to DRAM, T3 is followed by three TI (idle) cycles. 1-27 ISD-T360SB 1—HARDWARE Figure 1-29: Codec Short Frame Timing tCTp tCTp tCTp tCTp 4xtCTp tCTp tCTp tCTp tCTp 4xtCTp tCTp tCTp tCTp tCTp 4xtCTp tCTp tCTp tCTp tCTp 4xtCTp tCTp tCTp tCTp tCTp CTTL CCLK tCCLKia tCCLKa tCCLKh tCCLKh CFS0/ CFS1 tFSia tFSa tFSh tFSh BIT 7 CDOUT tCDOv tCDOv tCDOh BIT 7 CDIN tCDIs NOTE: tCDIh The CCLK and CFS0 timing is shown for Master Mode only. For Slave Mode, see Figure 1-31. Figure 1-30: Codec Long Frame Timing tCTp tCTp tCTp tCTp 4xtCTp tCTp tCTp tCTp tCTp 4xtCTp tCTp tCTp tCTp tCTp 4xtCTp tCTp tCTp tCTp tCTp CTTL CCLK tCCLKia tCCLKa tCCLKh tCCLKh CFS0/ CFS1 tFSia tFSa tFSh tFSh BIT 7 CDOUT BIT 0 tCDOv tCDOh CDIN NOTE: 1-28 BIT 7 The CCLK and CFS0 timing is shown for Master Mode only. For Slave Mode, see Figure 1-31. Voice Solutions in Silicon™ ISD-T360SB 1—HARDWARE Figure 1-31: Slave Codec CCLK and CFS0 Timing tCCLKSp CCLK tCCLKSh tCCLKSl CFS0 tCFS0Ss tCFS0Sh Figure 1-32: MICROWIRE Transaction Timing—Data Transmitted to Output ISD 1-29 ISD-T360SB 1—HARDWARE Figure 1-33: MICROWIRE Transaction Timing—Data Echoed to Output Figure 1-34: Master MICROWIRE Timing 1-30 Voice Solutions in Silicon™ ISD-T360SB 1—HARDWARE Figure 1-35: Output Signal Timing for Port PB and MWRQST NOTE: This cycle may be either TI (Idle), T2, T3 or T3H. Figure 1-36: CLKIN Timing Figure 1-37: CTTL Timing ISD 1-31 ISD-T360SB 1—HARDWARE Figure 1-38: Reset Timing When Reset Is Not at Power-Up Figure 1-39: Reset Timing When Reset Is at Power-Up 1-32 Voice Solutions in Silicon™ ISD-T360SB 2—SOFTWARE Chapter 2ÑSOFTWARE The VoiceDSP software resides in the on-chip ROM. It includes DSP-based algorithms, system support functions and a software interface to hardware peripherals. 2.1 the processor detectors (VOX, constant energy, call progress tones and DTMF) are not active. In all other states, these detectors are active. (See the SDET and RDET commands for further details). SYSTEM OPERATION This section provides details of the system support functions and their principle operation. It is divided into the following subjects: • The State Machine IDLE This is the state from which most commands are executed. As soon as a command and all its parameters are received, the processor starts executing the command. • Command Execution • Event Handling • Message Handling • Tone Generation PLAY In this state a message is decompressed (unless stored in PCM format), and played back. • Initialization and Configuration • Power Down Mode (PDM) RECORD In this state a message is compressed (unless stored in PCM format) and recorded into the message memory. 2.1.1 THE STATE MACHINE The ISD-T360SB operates in two modes, normal mode (DTAD) and speakerphone mode. To change the mode use the Set Speakerphone Mode (SSM) command. The VoiceDSP processor functions as a state machine under each mode. It changes state either in response to a command sent by the microcontroller, after execution of the command is completed, or as a result of an internal event (e.g. memory full or power failure). For more information see “VoiceDSP PROCESSOR COMMANDS—QUICK REFERENCE TABLE” on page 2-21. The VoiceDSP states are desctibed below: RESET SYNTHESIS An individual word or a sentence is synthesized from an external vocabulary in this state. TONE_GENERATE In the TONE_GENERATE state, the VoiceDSP processor generates single or DTMF tones. MSG_OPEN The VoiceDSP processor either reads or writes 32 bytes to the message memory, or sets the message read/write pointer on a 32 byte boundary. The VoiceDSP processor is initialized to the RESET state after a full hardware reset by the RESET signal (See “RESETTING” on page 1-3). In this state ISD 2-1 ISD-T360SB 2.1.2 COMMAND EXECUTION A VoiceDSP processor command is represented by an 8-bit opcode. Some commands have parameters and some have return values. Commands are either synchronous or asynchronous. SYNCHRONOUS COMMANDS A synchronous command must complete execution before the microcontroller can send a new command (e.g. GMS, GEW). A command sequence begins when the microcontroller sends an 8-bit opcode to the processor, followed by the command’s parameters (if any). The VoiceDSP processor then executes the command and, if required, transmits a return value to the microcontroller. Upon completion, the processor notifies the microcontroller that it is ready to accept a new command. 2—SOFTWARE 2.1.3 EVENT HANDLING STATUS WORD The 16-bit status word indicates events that occur during normal operation. The VoiceDSP processor activates the MWRQST signal, to indicate a change in the status word. This signal remains active until the processor receives a GSW (Get Status Word) command. For detailed description of the Status Word and the meaning of each bit, see “GSW Get Status Word” on page 2-33. ERROR WORD The 16-bit error word indicates errors that occurred during execution of the last command. If an error is detected, the command is not processed; the EV_ERROR bit in the status word is set to 1, and the MWRQST signal is activated. ASYNCHRONOUS COMMANDS An asynchronous command starts execution in the background and notifies the microcontroller, which can send more commands while the current command is still running (e.g. R, P). After receiving an asynchronous command, such as P (Playback), R (Record), SW (Say Words) or GT (Generate Tone), the VoiceDSP processor switches to the appropriate state and executes the command until finished or a S (Stop) or PA (Pause) command is received from the microcontroller. When completed, the EV_NORMAL _END event is set and the processor switches to the IDLE state. ERROR HANDLING When the microcontroller detects the active MWRQST signal, it issues the GSW command, deactivating the MWRQST signal. Then, the microcontroller tests the EV_ERROR bit in the status word, and, if set, sends the GEW (Get Error Word) command to read the error word for details. For detailed description of the Error Word and the meaning of each bit, see “GEW Get Error Word” on page 2-30. “VoiceDSP PROCESSOR COMMANDS—QUICK REFERENCE TABLE” on page 2–21 displays all the processor commands, the valid source states in which these commands are valid, and the states resulting from the command. 2-2 Voice Solutions in Silicon™ ISD-T360SB 2—SOFTWARE 2.1.4 MESSAGE HANDLING A message is the basic unit on which most of the VoiceDSP commands operate. A VoiceDSP processor message, stored on a memory device (Flash or ARAM/DRAM), can be regarded as a computer file stored on a mass-storage device. The ISD-T360SB manages messages for a wide range of applications, which require different levels of DTAD functionality. The VoiceDSP processor features advanced memory-organization features such as multiple OutGoing Messages (OGMs), mailboxes, and the ability to distinguish between InComing Messages (ICMs) and OGMs. A message is created with either the R (Record) or the CMSG (Create Message) command. Once created, the message is assigned a timeand-day stamp and a message tag which is read by the microcontroller. The R command takes voice samples from the codec, compresses them, and stores them in the message memory. When a message is created with the CMSG command the data to be recorded is provided by the microcontroller, via the WMSG (Write Message) command and not through the codec. Here, the data is transferred directly to the message memory, and not compressed by the ISDT360SB voice compression algorithm. WMSG, RMSG (Read Message) and SMSG (Set Message Pointer) are message-data access commands used to store and read data to or from any location in the message memory (see “VoiceDSP PROCESSOR COMMANDS—QUICK REFERENCE TABLE” on page 2-21 for more details). Using these commands, the microcontroller utilizes messages for features such as a Telephone Directory and playing back (P command) or deleting (DM command) a message. Remove redundant data (e.g., trailing tones or silence) from the message tail with the CMT (Cut Message Tail) command. CURRENT MESSAGE The GTM (Get Tagged Message) command selects the current message. Most message handling commands (P, DM, RMSG), operate on the current message. Deleting the prevailing message does not cause a different message to become current; the current message is undefined. If you issue the GTM command to skip to the next message, the first message, newer than the just deleted message, becomes the current message. MESSAGE TAG Each message has a 2-byte message tag which used to categorize messages, and implement such features as OutGoing Messages, mailboxes, and different handling of old and new messages. The tag is created during the R (Record) command. Use the GMT (Get Message Tag) and SMT (Set Message Tag) commands to handle message tags. NOTE Message tag bits can only be cleared and are set only when a message is first created. This limitation, inherent in Flash memories, allows bits to be changed only from 1 to 0 (changing bits from 0 to 1 requires a special erasure procedure). However, the usual reason for updating an existing tag is to mark a message as old. This can be done when a message is first created by using one of the bits as a new/old indicator, setting the bit to 1 and later clearing it when necessary. The PA (Pause) and RES (Resume) commands suspend the P and R commands, respectively, and then resume them from where they were suspended. ISD 2-3 ISD-T360SB 2.1.5 TONE GENERATION The VoiceDSP processor generates DTMF tones and single-frequency tones from 300Hz to 3000Hz in increments of 100Hz. The ISD-T360SB tone generation conforms to the EIA-470-RS standard. Note, however, that value of some tunable parameters may need adjusting to meet the standard specifications since the energy level of generated tones depends on the analog circuits used. 1. 2. 3. Tune the DTMF_GEN_TWIST_LEVEL parameter to control the twist level of the generated DTMF tones. Use the VC (Volume Control) command, and tune the TONE_GEN_LEVEL parameter, to control the energy level at which these tones are generated. Use the GT (Generate Tone) command to specify the DTMF tones, and the frequency at which single tones are generated. Refer to table 2-5, VC command and GT command of the Command Description section for furhter details of the relevant tunable parameters and commands. NOTE 2-4 The DTMF detector performance is degraded during tone generation, especially if the frequency of the generated tone is close to the frequency of one of the DTMF tones. 2—SOFTWARE 2.1.6 INITIALIZATION AND CONFIGURATION Use the following procedures to initialize the VoiceDSP processor: NORMAL INITIALIZATION Reset the VoiceDSP processor by activating the RESET signal. (See “RESETTING” on page 1-3.) 1. Issue a CFG (Configure VoiceDSP processor) command to change the configuration according to your environment. 2. Issue an INIT (Initialize System) command to initialize the VoiceDSP firmware. 3. Issue a series of TUNE commands to adjust the VoiceDSP processor to the requirements of your system. TUNABLE PARAMETERS The VoiceDSP processor can be adjusted to the specific system’s requirements using a set of tunable parameters. These parameters are set to their default values after reset and can be later modified with the TUNE command. By tuning these parameters, you can control various aspects of the VoiceDSP processor’s operation, such as silence compression, tone detection, and no-energy detection. Tables 2-4 to 2-11 of the Command Description section describe all the tunable parameters in detail. Voice Solutions in Silicon™ ISD-T360SB 2—SOFTWARE 2.1.7 POWER-DOWN MODE The PDM (Go To Power-Down Mode) command switches the ISD-T360SB to power-down mode. The purpose of the PDM command is to save power during buttery operation, or for any other power saving cause. During power-down mode only basic functions, such as ARAM/DRAM refresh and time and date update, are active (for more details refer to POWER-DOWN MODE description on page 1-4). This PDM command may only be issued when the processor is in the IDLE mode (for an explanation of the ISD-T360SB states, see “Command Execution” on page 2–2). If it is necessary to switch to power-down mode from any other state, the controller must first issue a S (Stop) command to switch the processor to the IDLE state, and then issue the PDM command. Sending any command while in power-down mode resets the VoiceDSP processor detectors, and returns it to normal operation mode. NOTE 2.2 Entering or exiting power-down mode can distort the real-time clock by up to 500 µs. Thus, to maintain the accuracy of the realtime clock, enter or exit the power-down mode as infrequently as possible. PERIPHERALS This section provides details of the peripherals interface support functions and their principle operation. It is divided into the following subjects: • Microcontroller Interface (Slave MICROWIRE) • Memory Interface • Codec Interface ISD 2.2.1 MICROCONTROLLER INTERFACE MICROWIRE/PLUS™ is a synchronous serial communication protocol minimizes the number of connections, and thus the cost, of communicating with peripherals. The VoiceDSP MICROWIRE interface implements the MICROWIRE/PLUS interface in slave mode, with an additional ready signal. It enables a microcontroller to interface efficiently with the VoiceDSP processor application. The microcontroller is the protocol master and provides the clock for the protocol. The VoiceDSP processor supports clock rates of up to 400 KHz. This transfer rate refers to the bit transfer; the actual throughput is slower due to byte processing by the VoiceDSP processor and the microcontroller. Communication is handled in bursts of eight bits (one byte). In each burst the VoiceDSP processor is able to receive and transmit eight bits of data. After eight bits have been transferred, an internal interrupt is issued for the VoiceDSP processor to process the byte, or to prepare another byte for sending. In parallel, the VoiceDSP processor sets MWRDY to 1, to signal the microcontroller that it is busy with the byte processing. Another byte can be transferred only when the MWRDY signal is cleared to 0 by the VoiceDSP processor. When the VoiceDSP processor transmits data, it expects to receive the value 0xAA before each transmitted byte. The VoiceDSP processor reports any status change by clearing the MWRQST signal to 0. If processor command’s parameter is larger than one byte, the microcontroller transmits the Most Significant Byte (MSB) first. If a return value is larger than one byte, the VoiceDSP processor transmits the MSB first. The following signals are used for the interface protocol. Input and output are relative to the VoiceDSP processor. 2-5 ISD-T360SB INPUT SIGNALS MWDIN MICROWIRE Data In. Used for input only, for transferring data from the microcontroller to the VoiceDSP processor. MWCLK MICROWIRE Clock. Serves as the synchronization clock during communication. One bit of data is transferred on every clock cycle. The input data is available on MWDIN and is latched on the clock rising edge. The transmitted data is output on MWDOUT on the clock falling edge. The signal should remain low when switching MWCS. MWCS MICROWIRE Chip Select. The MWCS signal is cleared to 0, to indicate that the VoiceDSP processor is being accessed. Setting MWCS to 1 causes the VoiceDSP processor to start driving MWDOUT with bit 7 of the transmitted value. Setting the MWCS signal resets the transfer-bit counter of the protocol, so the signal can be used to synchronize between the VoiceDSP processor and the microcontroller. To prevent false detection of access to the VoiceDSP processor due to spikes on the MWCLK signal, use this chip select signal, and toggle the MWCLK input signal, only when the VoiceDSP processor is accessed. OUTPUT SIGNALS MWDOUT MICROWIRE Data Out. Used for output only, for transferring data from the VoiceDSP processor to the microcontroller. When the VoiceDSP processor receives data it is echoed back to the microcontroller on this signal, unless the received data is 0xAA. In this case, the VoiceDSP processor echoes a command’s return value. 2-6 2—SOFTWARE MWRDY MICROWIRE Ready. When active (0), this signal indicates that the VoiceDSP processor is ready to transfer (receive or transmit) another byte of data. This signal is set to 1 by the VoiceDSP processor after each byte transfer has been completed. It remains 1, while the VoiceDSP processor is busy reading the byte, writing the next byte, or executing the received command (after the last parameter has been received). MWRDY is cleared to 0 after reset. For proper operation after a hardware reset, this signal should be pulled up. MWRQST MICROWIRE Request. When active (0), this signal indicates that new status information is available. MWRQST is deactivated (set to 1), after the VoiceDSP processor receives a GSW (Get Status Word) command from the microcontroller. After reset, this signal is active (0) to indicate that a reset occurred. MWRQST, unlike all the signals of the communication protocol, is an asynchronous line that is controlled by the VoiceDSP firmware. SIGNAL USE IN THE INTERFACE PROTOCOL After reset, both MWRQST and MWRDY are cleared to 0. The MWRQST signal is activated to indicate that a reset occurred. The EV_RESET bit in the status register is used to indicate a reset condition. The GSW command should be issued after reset to verify that the EV_RESET event occurred, and to deactivate the MWRQST signal. While the MWCS signal is active (0), the VoiceDSP processor reads data from MWDIN on every rising edge of MWCLK. VoiceDSP processor also writes every bit back to MWDOUT. This bit is either the same bit which was read from MWDIN (in this case it is written back as a synchronization echo after some propagation delay), or it is a bit of a value the VoiceDSP processor transmits to the microcontroller (in this case it is written on every falling edge of the clock). Voice Solutions in Silicon™ ISD-T360SB 2—SOFTWARE When a command has more than one parameter/return-value, the parameters/return-values are transmitted in the order of appearance. If a parameter/return-value is more than one byte long, the bytes are transmitted from the most significant to the least significant. Figure 1-32 and Figure 1-33 illustrate the sequence of activities during a MICROWIRE data transfer between VoiceDSP and the microcontroller. The MWRDY signal is used as follows: INTERFACE PROTOCOL TIME-OUTS 1. Active (0) MWRDY signals the microcontroller that the last eight bits of data transferred to/from the voice module were accepted and processed (see below). 2. The MWRDY signal is deactivated (set to 1 by the VoiceDSP processor) after 8-bits of data were transferred to/from the VoiceDSP processor. The bit is set following the falling edge of the eighth MWCLK clock-cycle. 3. 4. 5. The MWRDY signal is activated (cleared to 0) by the VoiceDSP processor when it is ready to receive the first parameter byte (if there are any parameters) and so on till the last byte of parameters is transferred. An active MWRDY signal after the last byte of parameters indicates that the command was parsed and (if possible) executed. If that command has a return value, the microcontroller must read the value before issuing a new command. When a return value is transmitted, the MWRDY signal is deactivated after every byte, and activated again when the VoiceDSP processor is ready to send another byte, or to receive a new command. The MWRDY signal is activated (cleared to 0) after reset, and after a protocol timeout. (See “INTERFACE PROTOCOL TIMEOUTS” ) The MWRQST signal is used as follows: 1. The MWRQST signal is activated (cleared to 0), when the status word is changed. 2. The MWRQST signal remains active (0), until the VoiceDSP processor receives a GSW command. ISD Depending on the VoiceDSP processor’s state, if more than 100 milliseconds elapse between the assertion of the MWRDY signal and the transmission 8th bit of the next byte pertaining to the same command transaction, a time-out event occurs, and the VoiceDSP processor responds as follows: 1. Sets the error bit in the status word to 1. 2. Sets the EV_TIMEOUT bit in the error word to 1. 3. Activates the MWRQST signal (clears it to 0). 4. Activates the MWRDY signal (clears it to 0). 5. Waits for a new command. (After a timeout occurs, i.e., the microcontroller received MWRQST during the command transfer, or result reception, the microcontroller must wait at least four milliseconds before issuing the next command.) ECHO MECHANISM The VoiceDSP processor echoes back to the microcontroller all the bits received by the VoiceDSP processor. Upon detection of an error in the echo, the microcontroller should stop the protocol clock, which eventually causes a time-out error (i.e., ERR_TIMEOUT bit is set in the error word). NOTE When a command has a return value, the VoiceDSP processor transmits bytes of the return value instead of the echo value. The VoiceDSP processor transmits a byte as an echo when it receives the value 0xAA from the microprocessor. Upon detection of an error the VoiceDSP processor activates the MWRQST signal, and sets the ERR_COMM bit in the error word. 2-7 ISD-T360SB 2.2.2 MEMORY INTERFACE DEVICE NUMBER AND TYPE The VoiceDSP processor supports various types of Flash memory and ARAM/DRAM devices. Up to four devices may be connected to the VoiceDSP, where all the connected devices must be of the same type. Each memory device may be of 4Mbit, 8Mbit or 16Mbit; thus a total of 64Mbit non-volatile memory may be connected for message storage (up to 4 hours of voice recording). See “MEMORY INTERFACE” on page 1-6, for detailed description of the supported Flash and ARAM/DRAM devices and the hardware connectivity. Use the CFG command to define the type and number of installed memory devices (see “CFG Configure VoiceDSP config_value” on page 225). MEMORY DEVICE SIZE The memory manager handles the memory devices in basic units of 4Kbyts blocks. This approach is defined due to the nature of Flash devices where the basic unit that can be erased is a 4Kbytes block. This constraint is not relevant for ARAM/DRAM devices, but the concept is maintained for simplicity and consistency. Memory blocks cannot be shared by different voice messages. Therefore, the maximum number of messages per memory device, equals to the number of memory blocks minus one (one block per device is used for memory management). 2—SOFTWARE PRODUCTION LINE TESTING In many cases it is desired to test the ISD-T360SB in the production line as part of the whole application. Usually in these cases, the testing time is an important factor and should be minimized as possible. The initialization time of the memory devices is significant and should be avoided during production (Refer to Table 1-4). Therefore, a dedicated parameter is defined in order to allow a production line testing while using a small part of the real connected memory size. It should be noted that in case of power failure during the production line testing, the connected memory devices should be replaced, and the process should be repeated. Refer to parameter index 63, in Table 2-10, for further explanation of the production line testing. ARAM QUALITY ARAM is an Audio Grade RAM device, which implies that some percentage of the ARAM bits are defected and their content is undefined. Unlike Flash devices, where the defected bits can be mapped out, in case of ARAM specific bits cannot be mapped out; only memory blocks can be mapped out. Therefore, it should be noted that using ARAM as a voice storage device, will result in audible distortions. It is the user responsibility to define the maximum allowed bad nibbles (4 bits) in a memory block. If the number of bad nibbles exceeds the defined limitation, the specific block is mapped out and is not used for message recording. Refer to tunable parameter index 64, in Table 2-10, for further details of the ARAM quality level definition. The size of the connected memory devices, is defined by the number of memory blocks in each device. Refer to tunable parameter index 62, in Table 2-10, for detailed description of the available number of blocks for Flash and ARAM/ DRAM devices. 2-8 Voice Solutions in Silicon™ ISD-T360SB 2—SOFTWARE 2.2.3 CODEC INTERFACE SUPPORTED FUNCTIONALITY The VoiceDSP processor supports analog and digital telephony in various configurations. For analog telephony the VoiceDSP operates in master mode, where it provides the clock and the synchronization signals. It supports a list of single channel and dual channel codecs, as listed in Table 1-7. For digital telephony the VoiceDSP operates in slave mode, where the control signals are provided by an external source. The codec interface is designed to exchange data in short frame format as well as in long frame format. The channel width may be either 8 bits (u-Law format or A-Law format), or 16 bits (linear format). In slave mode the clock may be divided by two, if required (two bit rate clock mode). DATA CHANNELS TIMING Especially in digital telephony, but also in analog telephony when speakerphone is connected, the channels data may be delayed from the synchronization signal by variable number of clock cycles. In order to allow full flexibility of the data delay relative to the synchronization signal, and the delay between the two synchronization signals, a set of registers is provided. setting the delay parameters of these registers defines the exact timing of all the codec interface signals. Refer to tunable parameters index 65 to index 69, in Table 2-11, for detailed description of the delay registers and their significance. 2.3 ALGORITHM FEATURES The VoiceDSP support up to 2 voice channels, where the line should be connected as channel 0 (in master mode or in slave mode - depends on the configuration), and the speakerphone (speaker and microphone) should be connected as channel 1 or as channel 2, depends on the configuration (channel 1 and channel 2 are always connected as master). This section provides details of the VoiceDSP algorithms and their principle operation. It is divided into the following subjects: See “The Codec Interface” on page 1-13, for detailed description of the supported codec devices and the hardware connectivity. • Speakerphone • VCD (Voice Compression and Decompression) • DTMF Detection • Tone and Energy Detection (Call Progress) • Speech Synthesis Use the CFG command to define the codec mode (master or slave), the data frame format (short or long), the channel width (8 bits or 16 bits), the clock bit rate (single or dual) and the number and type of codecs (one or two, single channel or dual channel). See “CFG Configure VoiceDSP config_value” on page 2-25. ISD 2-9 ISD-T360SB 2.3.1 VCD (VOICE COMPRESSION AND DECOMPRESSION) The VoiceDSP processor implements a state of the art VCD algorithm of the CELP family. The algorithm provides 3 compression rates that can be selected dynamically (actually, the algorithm supports more compression rates). PCM recording (no compression) is also provided. The lowest compression rate of 5.3 Kbit/s enables about 30 minutes of recording on an 8-Mbit device (depending on the relative silence period). The mid-quality compression rate of 9.9 Kbit/s provides about 16 minutes of voice recording time. The highest compression rate of 16.8 Kbit/s, the highest quality recording, stores up to 10 minutes on a 8-Mbit device. For detailed information about recording times refer to table 1-5. Before recording each message, the microcontroller selects one of the three compression rates, or PCM recording, with the compression_rate parameter of the R (Record) command. During message playback the VoiceDSP processor reads this one byte parameter and selects the appropriate speech decompression algorithm. IVS vocabularies can be prepared in either of the three compression rates, or in PCM format, using the IVS tool. All messages in a single vocabulary must be recorded using the same algorithm. (See the IVS User’s Guide for more details). During speech synthesis, the VoiceDSP processor automatically selects the appropriate speech decompression algorithm. SILENCE COMPRESSION A Voice Activity Detector (VAD) is used in order to detect periods of silence during the compression of the recorded message. Silence is treated differently than normal voice by the compression algorithm. It is compressed to about 1.0 Kbit/ s. The compressed silence contains data that allows to generate comfort noise during message playback. The comfort noise generation is important because the human ear is not used to “real” silence while listening to messages. Various tunable parameters are available in order to optimally tune the VAD. The silence com- 2-10 2—SOFTWARE pression may be turned Off, though it is planned to remain On continuously. For more details refer to table 2-4 of the Command Description section. NOTE The silence compression should be turned Off when ARAM devices are used for voice storage. Otherwise, unpredictable results are expected during message playback. SW AGC A SoftWare Automatic Gain Control (SW AGC) algorithm is activated with the compression module in order to regulate the input signal to a dynamic range that will provide higher compression quality. The algorithm senses the energy level and updates the signal gain in order to amplify low energy signals and to avoid signal saturation. The SW AGC feature eliminates the need for an external HW AGC, thus reducing hardware costs and complexity. Hardware Gain Control may be used to avoid signal saturation prior to sampling the signal. A tunable parameter determines the maximum allowed gain for the SW AGC algorithm. The SW AGC may be turned Off, though it is planned to remain On continuously. For more details refer to table 2-4 of the Command Description section. VARIABLE SPEED PLAYBACK This feature increases or decreases the speed of messages and synthesized messages during playback. Use the SPS (Set Playback Speed) to set the speed of message playback. The new speed applies to all recorded messages and synthesized messages (only if synthesized using IVS), until changed by another SPS command. If this command is issued while the VoiceDSP processor is in the PLAY state, the speed also changes for the message currently being played. The speedup / slowdown algorithm is designed to maintain the pitch of the original speech. This approach provides the same speech tone while playback speed varies. Voice Solutions in Silicon™ ISD-T360SB 2—SOFTWARE PCM RECORDING The VoiceDSP is capable of recording data in PCM format (that is the original samples format either in 8 bits u-Law format, 8 bit A-law format or 16 bits linear format). The PCM data uses more storage space, but it provides the highest quality for OGM, music-on-hold or IVS data. The PCM recording may be selected as one of the available compression rates during the R command (compression_rate = 0). Silence Compression and variable Speed Playback are not feasible during PCM recording and playback since this feature skips the compression algorithm. 2.3.2 active throughout the operation of the VoiceDSP processor. Detection can be configured using the SDET (Set Detectors Mask) command, which controls the reporting of the occurrence of tones, and the RDET (Reset Detectors) command which resets the detectors. The accuracy of the tone length, as reported by the tone detectors, is ±10 ms. DTMF detection may be reported at the starting point, ending point, or both. The report is made through the status word (for further details, see GSW command). For further details about tunable parameters refer to table 2-6 of the Command description section. The DTMF detector performance, as measured on the line input using an ISD-DS360-DAA board, is summarized below (see Table 2-1). DTMF DETECTION The VoiceDSP processor detects DTMF, which enables remote control operations. Detection is Table 2-1: DTMF Detector Performance1 Play/IVS Synthesis Record/Idle Detection Sensitivity Performance depends on the message being played.2 −34 dBm Accepted DTMF Length3 >50 ms >40 ms Frequency Tolerance ±1.5% ±1.5% S/N Ratio 12 dB 12 dB >50 ms >45 ms 8 dB 8 dB 4 dB or 8 dB 4 dB or 8 dB Minimum Spacing4 Normal Twist Reverse 1. 2. 3. 4. 5. 6. ISD Twist5 Performance depends on the DAA design. For reliable DTMF detection: - A hardware echo-canceler, that attenuates the echo by at least 6 dBm, is required during playback. - The HW AGC, if present, must be disabled during playback. Performance with echo canceler is 10 dB better than without echo canceler. For a silent message, Detection sensitivity is −34 dBm, with echo canceler. Tune parameters 60 and 61 may improve DTMF detection sensitivity. for more details refer to the parameters description in Table 2-6. The accuracy of reported DTMF tones is ±10 ms. If the interval between two consecutive identical DTMF tones is less than, or equal to, 20 ms, the two are detected as one long DTMF tone. If the interval between two consecutive identical DTMF tones is between 20 ms and 45 ms, separate detection is unpredictable. Determined by the DTMF_REV_TWIST tunable parameter value. 2-11 ISD-T360SB DTMF SW AGC In order to remove the linkage between the HW AGC and the detection level of the DTMF detector, two new tunable parameters are added. These tunable parameters define the gain of the SW AGC for DTMF signals. DTMF_DET_AGC_IDLE - SW AGC for DTMF detection in Idle and Record states. When incrementing this tunable by 1, the dynamic range is increased by 3 dB. DTMF_DET_AGC_PLAY - SW AGC for DTMF detection in Play and Tone_Generate states. When incrementing this tunable by 1, the dynamic range is increased by 3 dB. 2—SOFTWARE 2.3.3 TONE AND ENERGY DETECTION (CALL PROGRESS) The VoiceDSP processor detects busy and dial tones, constant energy level, and no-energy (VOX). This enables call progress tracking. Detection is active throughout the operation of the VoiceDSP processor. Detection can be configured using the SDET (Set Detectors Mask) command, which controls the reporting of the occurrence of tones, and the RDET (Reset Detectors) command which resets the detectors. The accuracy of the tone length, as reported by the tone detectors, is ±10 ms. TUNABLE PARAMETERS ECHO CANCELLATION Echo cancellation is a technique used to improve the performance of DTMF detection during speech synthesis, tone generation, and OGM playback. For echo cancellation to work properly, HW AGC must not be active in parallel. Thus, to take advantage of echo cancellation, the microcontroller must control the HW AGC, if exists, (i.e., disable the HW AGC during PLAY, SYNTHESIS and TONE_GENERATE states and enable it again afterwards). If HW AGC can not be disabled, do not use echo cancellation. Tunable parameters control the detection of busy and dial tones, constant energy level (in the frequency range 200–3400Hz), and no-energy. These parameters should be tuned to fit the system hardware. In addition, changes may be required to the tunable parameters according to the setting (On or Off) of the HW Automatic Gain Control (HW AGC), if exists. For more information refer to tables 2-7, 2-8 of the Command Description section. The microcontroller should use the CFG command to activate/deactivate echo cancellation. NOTE 2-12 Normally, a HW AGC is not required with The ISD-T360SB, since SW AGC is active for the VCD algorithm, DTMF detection and the speakerphone module. Voice Solutions in Silicon™ ISD-T360SB 2—SOFTWARE Figure 2-1: Busy and Dial-Tone Band-Pass Filter Frequency Response 0 Magnitude dB –10 –20 –30 –40 0 200 400 600 800 1000 1200 1400 1600 1800 2000 Frequency (Hz) BUSY AND DIAL TONES termined by the BUSY_DET_TONE_TYPE tunable parameter. Busy and dial-tone detectors work with a bandpass filter that limits the frequency range in which tones can be detected to 0–1100Hz. Figure 1-1 shows the frequency response of this band-pass filter. BUSY_DET_TONE_TYPE specifies the type of cadences that are supported. Legal values are: The design of the busy-tone detector allows very high flexibility in detecting busy tones with varying cadences. Two cadences only Three cadences only Both two and three cadences. The tunable parameters are divided into five sets: The acceptance criteria for two cadences: 1. 2. Busy Tone On-time and Off-time Range Specification: [E1–E3] < BUSY_DET_DIFF_THRESHOLD and [S1–S3] < BUSY_DET_DIFF_THRESHOLD BUSY_DET_MIN_ON_TIME BUSY_DET_MIN_OFF_TIME BUSY_DET_MAX_ON_TIME BUSY_DET_MAX_OFF_TIME The acceptance cadences: BUSY_DET_VERIFY_COUNT determines the number of On/Off cadences that detector should detect before reporting busy tone presence. for three [E1–E4] < BUSY_DET_DIFF_THRESHOLD and [S1–S4] < BUSY_DET_DIFF_THRESHOLD Busy Tone Cadence Control Specification BUSY_DET_VERIFY_COUNT BUSY_DET_TONE_TYPE BUSY_DET_DIFF_THRESHOLD criteria 3. Busy and Dial Tone Energy Thresholds TONE_DET_ON_ENERGY THRESHOLD TONE_DET_OFF_ENERGY THRESHOLD 4. Busy Detection Time BUSY_DET_MIN_TIME BUSY_DET_DIFF_THRESHOLD describes the maximum allowed difference between two compared On or Off periods, as de- ISD 2-13 ISD-T360SB 2—SOFTWARE Figure 2-2: Busy-Tone Detector—Default Cadence Specification E1 E2 E3 S1 [E1 S2 S3 – E3] < 100 ms [S1 – S3] < 100 ms 100 < Ei < 1680 ms 70 < Si < 1220 ms CONSTANT ENERGY 2.3.4 The constant-energy detector reports the presence of constant energy in the range 200Hz to 3400Hz. It is intended to detect both white and pink noise and can be used to detect line disconnection during recording. The speakerphone feature lets the user communicate through a telephone line, using the unit’s speaker and the microphone instead of its handset. The speakerphone processes signals sent from the line to the speaker, and from the microphone to the line. It also performs the necessary switching, attenuation and echo cancellation on the signals present on the line/speaker. It is recommend to use the constant energy mechanism in conjunction with the no-energy (VOX) mechanism. The following tunable parameters control the operation of the constant-energy detector: CONST_NRG_DET_TIME_COUNT CONST_NRG_DET_TOLERANCE_TIME CONST_NRG_DET_LOW_THRESHOLD CONST_NRG_DET_HIGH_THRESHOLD NO ENERGY (VOX) The no-energy detector reports when the energy in the frequency range 200Hz to 3400Hz remains below a preprogrammed threshold for a preprogrammed time-out. A programmable tolerance is allowed. It is recommend to use the no-energy (VOX) mechanism in conjunction with the constant-energy mechanism. The following tunable parameters control the operation of the no-energy (VOX) mechanism: VOX_DET_ENERGY_THRESHOLD VOX_DET_TIME_COUNT VOX_DET_TOLERANCE_TIME 2-14 FULL-DUPLEX SPEAKERPHONE The ISD-T360SB speakerphone is simple to use; it requires no special hardware or training for the echo cancelers. The gain control is fully digital, which eliminates the need for analog gain control hardware. The speakerphone features two types of echoes, the electrical echo (line or circuit) and the acoustic echo. The electrical echo is a result of an imperfect impedance match between the 4to 2-wire interface (hybrid) and the line impedance. The electrical echo, relatively short term, has a transfer function that varies slowly. The second echo, the acoustic echo, is a line impedance returning from the speaker to the microphone. This echo is relatively long term, and its transfer function may vary quite quickly if anyone, or anything, moves in the room. Both echoes must be canceled to achieve a highquality hands-free system. For more details of the speakerphone tunable parameters refer to table 2-9 of the Command Description section. Voice Solutions in Silicon™ ISD-T360SB 2—SOFTWARE SPEAKERPHONE TERMINOLOGY Hold Send Path During Hold mode interrupts from both codecs are stopped. Neither side can hear each other. The signal path from the microphone (near-end speaker) to the line (far-end listener). The microphone is the input port, and line-out is the output port of this signal path. Restart The signal path from the line (far-end speaker) to the loudspeaker (near-end listener). The line-in is the input port, and the speaker is the output port for this signal path. In Restart mode the speakerphone re-initializes itself to the last speakerphone mode (full-duplex, transparent or mute). This mode should be used to resume the speakerphone operation after Hold mode or when there is a significant change in the environmental conditions (e.g., parallel pickup) that may affect the speakerphone quality. AEC Transparent Acoustic Echo Controller. The part in the speakerphone algorithm that controls the echo in the sendpath. While in Transparent mode, the speakerphone works in full-duplex mode but without echo cancellation. EEC Samples from the microphone are transferred to the line, and samples from the line are transferred to the speaker, with no processing. This mode should be used only for tuning and testing the system. Receive Path Electric Echo Controller. The part in the speakerphone algorithm that controls the echo in the receive path. Listen SPEAKERPHONE MODES OF OPERATION Full-Duplex (ON) The speakerphone works in full-duplex mode, meaning both parties can speak and hear each other simultaneously. In this mode both the acoustic and electric echo controllers are active. The VoiceDSP processor tone detectors are not active in this mode. In Listen mode the line is audible on the speaker, and the processor tone detectors are active. During Listen mode, dialing with the GT command and call progress is possible, since the busy and dial tone detectors are active. The following pseudo-code demonstrates how to make a call from speakerphone mode: Mute In this mode of operation, the speakerphone generates silence to the line. The near-end listener can hear the far-end speaker but not vice versa. Tone detectors are not active. ISD 2-15 ISD-T360SB 2—SOFTWARE Figure 2-3: Speakerphone Pseudo Code Representation while () { EV = wait_event() case EV of: skpr_button_pressed: if (speakerphone_on) { SSM 0 // Put VoiceDSP in idle mode first_digit = TRUE deactivate_digit_timeout_event() else SSM 1 // Put VoiceDSP in full-duplex speakerphone mode digit_pressed: if (first_digit) { SSM 4 // Enter LISTEN mode first_digit = FALSE } GT <dtmf_of_digit> // Dial the digit S // Stop. Note that after the S command // the VoiceDSP is still in speakerphone mode enable_digit_timeout_event() // To "guess" when dialing is completed. digit_timeout_event: SSM 1 // Dialing is completed, Go back to full-duplex mode deactivate_digit_timeout_event() } 2.3.5 SPEECH SYNTHESIS Speech synthesis is the technology used to create messages out of predefined words and phrases stored in a vocabulary. There are two kinds of predefined messages: fixed messages (voice menus in a voice-mail system) and programmable messages (time-andday stamp, or the You have n messages announcement in a DTAD). A vocabulary includes a set of predefined words and phrases, needed to synthesize messages in any language. Applications which support more than one language require a separate vocabulary for each language. INTERNATIONAL VOCABULARY SUPPORT (IVS) IVS is a mechanism by which the VoiceDSP processor utilizes several vocabularies stored on an external storage device. IVS enables the ISDT360SB to synthesize messages with the same meaning, but in different languages, from separate vocabularies. IVS Features • Multiple vocabularies stored on a single storage device. • Plug-and-play. The same microcontroller code is used for all languages. • Synthesized and recorded messages use the same voice compression algorithm to achieve equal quality. • Argumented sentences. (For example: You have <n> messages.) 2-16 Voice Solutions in Silicon™ ISD-T360SB 2—SOFTWARE • Auto-synthesized time-and-day stamp (driven by the VoiceDSP processor’s clock). • Support for various language and sentence structures: – One versus many. (For example: You have one message versus You have two messages.) – None versus many. (For example: You have no messages versus You have two messages.) – Number synthesis (English—Eighty versus French—Quatre-vingt). – Word order (English—Twenty-one versus German—Einundzwanzig). – Days of the week (Monday through Sunday versus Sunday through Saturday). VOCABULARY DESIGN There are several issues, sometimes conflicting, which must be addressed when designing a vocabulary. Vocabulary Content If memory space is not an issue, the vocabulary could contain all the required sentences, each recorded separately. If memory space is a concern, the vocabulary must be compact; it should contain the minimum set of words and phrases required to synthesize all the sentences. The least memory is used when phrases and words that are common to more than one sentence are recorded only once, and the IVS tool is used to synthesize sentences out of them. A good combination of sentence quality and memory space is achieved if you take the “compact” approach, and extend it to solve pronunciation problems. For example, the word twenty is pronounced differently when used in the sentences You have twenty messages and You ISD have twenty-two messages. To solve this problem, words that are pronounced differently should be recorded more than once, each in the correct pronunciation. Vocabulary Recording When recording vocabulary words, there is a compromise between space and quality. On one hand, the words should be recorded and saved in a compressed form, and you would like to use the best voice compression for that purpose. On the other hand, the higher the compression rate, the worse the voice quality. Another issue to consider is the difference in voice quality between synthesized and recorded messages (e.g., between time-and-day stamp and ICMs in a DTAD environment). It is more pleasant to the human ear to hear both messages have the same sound quality. Vocabulary Access Sometimes compactness and high quality are not enough. There should be a simple and flexible interface to access the vocabulary elements. Not just the vocabulary but the code to access the vocabulary should be compact. When designing for a multi-lingual environment, there are even more issues to consider. Each vocabulary should be able to handle languagespecific structures and designed in a cooperative way with the other vocabularies so that the code to access each vocabulary is the same. When you use the command to synthesize the sentence Monday. 12:30 P.M., you should not care in what language it is going to be played back. IVS VOCABULARY COMPONENTS This section describes the basic concept of an IVS vocabulary, its components, and the relationships between them. 2-17 ISD-T360SB Basic Concepts An IVS vocabulary consists of words, sentences, and special codes that control the behavior of the algorithm which VoiceDSP processor uses to synthesize sentences. Word Table The words are the basic units in the vocabulary. Create synthesized sentences by combining words in the vocabulary. Each word in the vocabulary is given an index which identifies it in the word table. Note that, depending on the language structures and sentences synthesized, you may need to record some words more than once in the vocabulary. For example, if you synthesize the sentences: you have twenty messages and you have twenty-five messages, the word twenty is pronounced differently. They should, therefore, be defined as two different words. Number Tables The number tables allow you to treat numbers differently depending on the context. Example 1: The number 1 can be announced as one as in message number one or as first as in first message. 2—SOFTWARE table is to make the microcontroller that drives the VoiceDSP processor independent of the language being synthesized. For example, if the Flash and/or ROM memory contain vocabularies in various languages, and the first sentence in each vocabulary means you have n messages, the microcontroller switches languages by issuing the following command to VoiceDSP processor: SV <storage_media>, <vocabulary_id> -Select a new vocabulary The microcontroller software is thus independent of the grammar of the language in use. The sentences consist of words, which are represented by their indices in the vocabulary. Sentence 0 All sentences but one are user defined. The VoiceDSP processor treats the first sentence in the sentence table (sentence 0) specially, to support time-and-day stamp. The processor assumes that the sentence is designed for both system time, and message time-and-day stamp announcement, and has one argument which is interpreted as follows: 0 System time is announced 1 The time-and-day stamp of the current message is announced. Example 2: The number 0 can be announced as no as in you have no messages or as oh as in monday, eight oh five A.M. Example 1: When the microcontroller sends the command: SAS 0, 0 A separate number table is required for each particular type of use. The number table contains the indices of the words in the vocabulary that are used to synthesize the number. Up to nine number tables can be included in a vocabulary. Example 2: When the microcontroller sends the command: SAS 0, 1 The system time and day is announced. The current message time-and-day stamp is announced. The following Figure 2-4 shows the interrelationship between the three types of tables. Sentence Table The sentence table describes the predefined sentences in the vocabulary. The purpose of this 2-18 Voice Solutions in Silicon™ ISD-T360SB 2—SOFTWARE Figure 2-4: The Interrelationship between the Word, the Number, and the Sentence Tables Sentence Table You have OPT_NUMBER Word Table CONTROL_SING MESSAGES five twenty You have messages message Number Table Control and Option Codes IVS Compiler The list of word indices alone cannot provide the entire range of sentences that the VoiceDSP processor is able to synthesize. IVS control and option codes send special instructions to control the speech synthesis algorithm’s behavior in the processor. The IVS compiler runs on MS-DOS (version 5.0 or later) and enables you to insert your own vocabulary, (i.e., basic words and data used to create numbers and sentences, as directories and files in MS-DOS). The IVS compiler then outputs a binary file containing that vocabulary. In turn, this information can be burned into an EPROM or Flash memory to be used by the VoiceDSP software. For example, if the sentence should announce the time of day, the VoiceDSP processor should be able to substitute the current day and time in the sentence. These control words do not represent recorded words, rather they instruct the processor to take special actions. NOTE The IVS data cannot be stored in EPROM when semi-parallel Flash is used (Samsung or Toshiba). THE IVS TOOL The IVS tool includes two utilities: 1. The DOS-based IVS Compiler 2. IVSTOOL for Windows. A Windows 3.1/95 based utility The tools help create vocabularies for the VoiceDSP processor. They take you from designing the vocabulary structure, through defining the vocabulary sentences, to recording the vocabulary words. ISD IVS Voice Compression Each IVS vocabulary can be compiled with either the 5.3 Kbit/s, the 9.9 Kbit/s or the 16.8Kbit/s voice compression algorithm, or in PCM format. Define the compression rate before compilation. The VoiceDSP processor automatically selects the required voice decompression algorithm when the SV command chooses the active vocabulary. 2-19 ISD-T360SB 2—SOFTWARE Graphical User Interface (GUI) The IVS package includes a Windows utility to assist the vocabulary designer to synthesize sentences. With this utility, you can both compose sentences and listen to them. HOW TO USE THE IVS TOOL WITH THE VOICEDSP PROCESSOR The IVS tool creates IVS vocabularies, and stores them as a binary file. This file is burnt into a ROM device or programmed into a Flash memory device using the INJ (Inject IVS) command. The VoiceDSP processor SO (Say One Word) command is used to select the required vocabulary. The SW (Say Words), SO, SS (Say Sentence) and SAS (Say Argumented Sentence) commands are used to synthesize the required word or sentence. The typical vocabulary-creation process is as follows: 1. Design the vocabulary. 2. Create the vocabulary files (as described in detail below). Use VISTULA for Windows 3.1/95 to simplify this process. 3. Record the words using any standard PC sound card and sound editing software, that can create .wav files. 4. Run the IVS compiler to compress the .wav files, and compile them and the vocabulary tables into an IVS vocabulary file. 5. Repeat steps 1 to 4 to create a separate IVS vocabulary for each language that you want to use. 6. Burn the IVS vocabulary files into a ROM (or Flash memory) device. Use the INJ (Inject IVS) command to program the data into a Flash device. NOTE The IVS data cannot be stored in EPROM when semi-parallel Flash is used (Samsung or Toshiba). Once the vocabulary is in place, the speech synthesis commands of the VoiceDSP processor can be used to synthesize sentences. Figure 2-5 shows the vocabulary-creation process for a single table on a ROM or Flash memory device. Figure 2-5: Creation of an IVS Vocabulary .wav File Editor .wav Files Compressed Files (.vcd) Number Tables PC + Sound Card IVS Compiler INJ IVS Vocabulary Command Files Sentence Table IVSTOOL for Windows .ini File ROM Programmer Flash ROM Editor 2-20 Voice Solutions in Silicon™ ISD-T360SB 2—SOFTWARE 2.4 VOICEDSP PROCESSOR COMMANDS—QUICK REFERENCE TABLE Table 2-2: Speech Commands Command Description Name S/A Opcode Hex Source State Result State Command Parameters Description Bytes Return Value Description Bytes CCIO S Configure Codec I/O 34 RESET, IDLE No change Config_value 1 None - CFG S Configure VoiceDSP 01 RESET No change Config_value 3 None - CMSG S Create Message 33 IDLE MSG_OPEN Tag, Num_of_ blocks 2+2 None - CMT S Cut Message Tail 26 IDLE No change Time_length 2 None - CVOC S Check Vocabulary 2B IDLE No change None - Test result 1 DM S Delete Message 0A IDLE No change None - None - DMS S Delete Messages 0B IDLE No change Tag_ref, Tag_mask 2+2 None - GCFG S Get Configuration Value 02 RESET, IDLE No change None - Version 1 GEW S Get Error Word 1B All states No change None - Error word 2 GI S Get Information item 25 PLAY, RECORD, No change SYNTHESIS, TONE_ GENERATE, IDLE Item 1 Item value 2 GL S Get Length 19 IDLE No change None - Message length 2 GMS S Get Memory Status 12 IDLE No change None - Recording time left 2 GMT S Get Message Tag 04 IDLE No change None - Message tag 2 GNM S Get Number of Messages 11 IDLE No change Tag_ref, Tag_mask 2+2 Number of messages 2 GSW S Get Status Word 14 All states No change None - Status word 2 GT A Generate Tone 0D IDLE TONE_ GENERATE Tone (single Tone or DTMF) 1 None - GTD S Get Time and Day 0E IDLE No change Time_day_ option 1 Time and day 2 GTM S Get Tagged Message 09 IDLE No change Tag_ref, Tag_mask,Dir Message found 1 GTUNE S Get Tunable Parameter 06 IDLE, RESET No change Index 1 Parameter_ value 2 INIT S Initialize System 13 RESET, IDLE IDLE None - None - INJ S Inject IVS data 29 IDLE No change N, byte1...byten 4+n None - MR S Memory Reset 2A RESET, IDLE No change None - None - P A Playback 03 IDLE PLAY None - None - ISD 2+2+1 2-21 ISD-T360SB 2—SOFTWARE Table 2-2: Speech Commands (Continued) Command Description Name S/A Opcode Hex Source State Result State Command Parameters Description Bytes Return Value Description Bytes PA S Pause 1C PLAY, RECORD, No change SYNTHESIS, TONE_GENERATE, IDLE* None - None - PDM S Go To PowerDown Mode 1A IDLE No change None - None - R A Record Message 0C IDLE RECORD Tag (message Tag), Compression_ rate 2+1 None - RDET S Reset Detectors 2C IDLE No change Detectors_ reset_mask 1 None - RES S Resume 1D PLAY, RECORD, No change SYNTHESIS, TONE_GENERATE IDLE* None - None - RMSG S Read Message 32 IDLE, MSG_OPEN MSG_OPEN None - Data 32 S S Stop 00 All states but RESET IDLE None - None - SAS A Say Argumented Sentence 1E IDLE SYNTHESIS Sentence_n, Arg 1+1 None - SB S Skip Backward 23 PLAY, IDLE* No change Time_length 2 None - SDET S Set Detectors Mask 10 IDLE No change Detectors_ mask 1 None - SE S Skip to End of Message 24 PLAY, IDLE* No change None - None - SETD S Set Time and Day 0F IDLE No change Time_and_ day 2 None - SF S Skip Forward 22 PLAY, IDLE* No change Time_length 2 None - SMSG S Set Message Pointer 30 IDLE, MSG_OPEN MSG_OPEN Num_of_ pages 2 None - SMT S Set Message Tag 05 IDLE Message_tag 2 None - SO A Say One Word 07 IDLE SYNTHESIS Word_number 1 None - SPS S Set Playback Speed 16 PLAY, SYNTHESIS, IDLE No change Speed 1 None - SS A Say Sentence 1F IDLE SYNTHESIS Sentence_n 1 None - 2-22 No change Voice Solutions in Silicon™ ISD-T360SB 2—SOFTWARE Table 2-2: Speech Commands (Continued) Command Description Name S/A Opcode Hex Source State Result State Command Parameters Description Bytes Return Value Description Bytes SV S Set Vocabulary Type 20 IDLE No change Type, Id 1+1 None - SW A Say Words 21 IDLE SYNTHESIS 1+N None - TUNE S Tune Parameters 15 IDLE, RESET No change Index, Parameter_ value 1+2 None - VC S Volume Control 28 PLAY, No change Vol_level SYNTHESIS, IDLE, (increment/ TONE_GENERAT decrement) E 1 None - WMSG S Write Message 31 IDLE, MSG_OPEN 32 None - N, word1...word n NOTE: MSG_OPEN Data * Command is valid in IDLE state, but has no effect. S = Synchronous command A= Asynchronous command Table 2-3: Speakerphone Commands Command Description Name S/A Opcode Hex Source State Result State Command Parameters Description Bytes Return Value Description Bytes GEW S Get Error Word 1B TONE_GENERAT No E, IDLE change None - Error word 2 GSW S Get Status Word 14 TONE_GENERAT No E, IDLE change None - Status word 2 GT A Generate Tone 0D IDLE TONE_GE NERATE Tone (Single Tone or DTMF) 1 None - RDET S Reset Detectors 2C IDLE No change Detectors_res et_mask 1 None - S S Stop 00 TONE_GENERAT IDLE E, IDLE None - None - SDET S Set Detectors Mask 10 IDLE No change Detectors_ mask 1 None - SSM S Set Speakerphon e Mode 2F IDLE No change Mode 1 None - VC S Volume Control 28 TONE_GENERAT No E, IDLE change Vol_level (increment/ decrement) 1 None - NOTE: ISD * Command is valid in IDLE state, but has no effect. S = Synchronous command A = Asynchronous command 2-23 ISD-T360SB 2.5 2—SOFTWARE COMMAND DESCRIPTION The commands are listed in alphabetical order. The execution time for all commands, when specified, includes the time required for the microcontroller to retrieve the return value, where appropriate. The execution time does not include the protocol timing overhead, i.e., the execution times are measured from the moment that the command is detected as valid until the command is fully executed. NOTE: Each command description includes an example application of the command. The examples show the opcode issued by the microcontroller and the response returned by the VoiceDSP processor. For commands which require a return value from the processor, the start of the return value is indicated by a thick vertical line. Configure Codec I/O config_value CCIO Configures the voice sample paths in various states. It should be used to change the default VoiceDSP processor configuration. It is relevant only when two codecs are used and speakerphonr is connected. The config_value parameter is encoded as follows: Bit 0 Loopback control. 0 Loopback disabled (default). 1 Loopback enabled. In the RECORD state, the input samples are echoed back, unchanged (i.e., no volume control), to the same codec. Bit 1 Codec input control. 0 Input is received via the line codec (default). 1 Input is received via the speakerphone codec. Bits 2–3 Codec output control. 2-24 00 In PLAY, IDLE, SYNTHESIS and TONE_GENERATE states, output is to both codecs. In RECORD state, output is to the non-input codec (no volume control). If the loopback control bit is set, output in RECORD state is to both codecs as well (default). 01 Output in all states is to the line codec. 10 Output in all states is to the speakerphone codec. 11 Output in all states is to both codecs. Voice Solutions in Silicon™ ISD-T360SB 2—SOFTWARE Bits 4–7 Reserved. Example CCIO 01 Byte sequence: Description: Microcontroller VoiceDSP Enable loopback 34 01 34 01 Configure VoiceDSP config_value CFG Configures the VoiceDSP processor in various hardware environments. It should be used to change the default processor configuration. The config_value parameter is encoded as follows: Bits 0–3 Memory type. 0000 No Memory (default). 0001 A/DRAM. 0010 Reserved. 0011 Toshiba Serial Flash. 0100 Samsung Semi-Parallel Flash. 0101 Toshiba Semi-Parallel Flash. 0110 Reserved. 0111 Reserved. Bits 4–5 Number of installed memory devices. 00 1 (Default) 01 2 10 3 11 4 Bit 6-14 Reserved. Bit 15 Echo Cancellation Control (for DTMF Detection). ISD 0 Echo cancellation off (default). 1 Echo cancellation is on during playback. 2-25 ISD-T360SB 2—SOFTWARE Echo cancellation improves the performance of DTMF detection during playback. Echo cancellation can be turned on only with a system that can disable HW AGC (if present) during playback. A system featuring HW AGC, that cannot be controlled by the microcontroller (i.e., disabled or enabled), must not turn on this bit. Bit 16 Clock bit rate (in Slave Mode only). 0 One bit rate clock (default). 1 Two bit rate clock. Bit 17 Codec configuration. 0 Short-frame format (default). 1 Long-frame format (guaranteed by design but not tested). Bits 18–19 Codec type. 00 16-bit linear (default). 01 µ-Law. 10 A-Law. Bit 20 Codec interface mode. 0 Master codec interface (default). 1 Slave codec interface. Bits 21–22 Number and type of codecs 00 One single codec (default). 01 Two single codecs. 10 One dual codec. 11 Reserved. The codecs should be connected as follows: Telephone line or equivalent - always connected as channel 0. Speaker and microphone - connected as channel 1 in case of one dual codec (not applicable in slave mode), connected as channel 2 in case of two single codecs. 2-26 Voice Solutions in Silicon™ ISD-T360SB 2—SOFTWARE Bit 23 Reserved. Example CFG 144013 Byte sequence: Description: CMSG 01 14 40 13 Microcontroller 01 14 40 13 VoiceDSP Configure the VoiceDSP to work with: Single codec in Slave Mode and A-Law compressed samples. Data in Short Frame format and Single Bit Rate interface. Two Serial Toshiba Flash devices. Echo Cancellation for DTMF detectiopn is On. Create Message tag num_of_blocks Creates a new message with a message tag tag, allocates num_of_blocks 4-Kbytes blocks for the new message, and sets the message pointer to the beginning of the message data. CMSG switches the VoiceDSP processor to the MSG_OPEN state. The memory space available for the message data is computed as follows: (127 x num_of_blocks – 2) x 32 bytes. Once a message is open (the processor is in the MSG_OPEN state), the message pointer can be set to any position on a page (32 bytes) boundary within the message with the SMSG command. Modify the message contents with the WMSG command, and read with the RMSG command. The microcontroller must issue an S command to close the message and switch the VoiceDSP processor to the IDLE state. If the memory is full, EV_MEMFULL is set in the status word and no message is created. If the memory is not full but there is insufficient memory and the processor cannot allocate more memory space, EV_MEMLOW is set in the status word and no message is created. Example CMSG 0101 0001 Byte sequence: Description: ISD 33 01 01 00 Microcontroller 01 33 01 01 00 VoiceDSP 01 Create a new message with a tag=0101, and allocate 1 block (4 Kbytes) for its data. 2-27 ISD-T360SB 2—SOFTWARE Cut Message Tail time_length CMT Cut time_length units, in 10 ms segments, off the end of the current message. The maximum value of time_length is 6550. In case of silence, cut-time accuracy is 0.1 to 0.2 seconds (depends on compression rate). NOTE If time_length is longer than the total duration of the message, the EV_NORMAL_END event is set in the status word and the message becomes empty but not deleted. Use the DM (Delete Message), or DMS (Delete Messages), command to delete the message. A compressed frame represents 21 ms of speech, thus the minimum meaningful parameter is 3, (i.e., a 30 ms cut). CMT 1 or CMT 2 have no effect.The CMT command can not be used on data messages Example CMT 02BC Byte sequence: Description: CVOC Microcontroller VoiceDSP 26 02 BC 26 02 BC Cut the last seven seconds of the current message. Check Vocabulary Checks (checksum) if the IVS data was correctly programmed to the ROM or Flash device. If the vocabulary data is correct the return value is 1. Otherwise the return value is 0. If the current vocabulary is undefined, ERR_INVALID is reported. Example CVOC Byte sequence: Microcontroller 2B AA 2B 01 Description: VoiceDSP Check the current vocabulary. The VoiceDSP processor responds that the vocabulary is OK. DM Delete Message Deletes the current message. Deleting a message clears its message tag. Deleting the current message does not cause a different message to become current. The current message is undefined. If, for example, you issue the GTM command to skip to the next message, the first message that is newer than the just deleted message is selected as the current message. The memory space released by the deleted message is immediately available for recording new messages. 2-28 Voice Solutions in Silicon™ ISD-T360SB 2—SOFTWARE Example DM Byte sequence: Description: Microcontroller VoiceDSP 0A 0A Delete current message. Delete Messages tag_ref tag_mask DMS Deletes all messages whose message tags match the tag_ref parameter. Only bits set in tag_mask are compared i.e., a match is considered successful if: message tag and tag_mask = tag_ref and tag_mask where and is a bitwise AND operation. After the command completes execution, the current message is undefined. Use the GTM command to select a message to be the current message. The memory space released by the deleted message is immediately available for recording new messages. Example DMS FFC2 003F Byte sequence: Description: Microcontroller VoiceDSP 0B FF C2 00 3F 0B FF C2 00 3F Delete all old incoming messages from mailbox Number 2 in a system where the message tag is encoded as follows: Bits 0–2: mailbox ID 8 mailboxes indexed: 0 to 7 Bit 3: new/old message indicator 0—Message is old 1—Message is new Bits 4–5: message type 00—ICM/memo 01—OGM 10—Call transfer message Bits 6–15: not used Note: the description of the tag is an example only. All bits of the tag are user-definable. GCFG Get Configuration Value Returns a sequence of one byte with the following information: Bits 0–7 Magic number, which specifies the VoiceDSP firmware version. ISD 2-29 ISD-T360SB 2—SOFTWARE Example GCFG Byte sequence: Microcontroller 02 AA 01 Description: VoiceDSP Get the VoiceDSP processor magic number. The VoiceDSP processor responds that it is Version 1. 02 GEW Get Error Word Returns the 2-byte error word. ERROR WORD The 16-bit error word indicates errors that occurred during execution of the last command. If an error is detected, the command is not processed; the EV_ERROR bit in the status word is set to 1, and the MWRQST signal is activated (set to 0). The GEW command reads the error word. The error word is cleared during reset and after execution of the GEW command. If errors ERR_COMMAND or ERR_PARAM occur during the execution of a command that has a return value, the return value is undefined. The microcontroller must still read the return value, to ensure proper synchronization. 15 9 Res 8 7 6 5 4 3 2 1 0 Res ERR_ INVALID ERR_ TIMEOUT ERR_ COMM Res ERR_ PARAM ERR_ COMMAND ERR_ OPCODE Res The bits of the error word are used as follows: ERR_OPCODE Illegal opcode. The VoiceDSP processor does not recognize the opcode. ERR_COMMAND Illegal command sequence. The command is not legal in the current state. ERR_PARAM Illegal parameter. The value of the parameter is out of range, or is not appropriate for the command. ERR_COMM Microcontroller MICROWIRE communication error. 2-30 Voice Solutions in Silicon™ ISD-T360SB 2—SOFTWARE ERR_TIMEOUT Time-out error. Depending on the VoiceDSP processor’s state, more than 100 milliseconds elapsed between the arrival of two consecutive bytes (for commands that have parameters). ERR_INVALID Command can not be performed in current context. Example GEW Byte sequence: Description: 1B AA AA Microcontroller 1B 00 02 VoiceDSP Get the VoiceDSP processor error word (typically sent after GSW when EV_ERROR is reported in the status word). The VoiceDSP processor responds: ERR_OPCODE: Get Information item GI Returns the 16-bit value specified by item from one of the internal registers of the VoiceDSP processor. item may be one of the following: 0 The duration of the last detected DTMF tone, in 10 ms units. The return value is meaningful only if DTMF detection is enabled, and the status word shows that a DTMF tone was detected. 1 The duration of the last detected busy tone in 10 ms units. 2 The energy level of the samples in the last 10 ms. 3 The energy level of the samples, in the last 10 ms, that are in the frequency range described in Figure 2-1. The return value is meaningful only if one of the tone detectors is enabled (bits 0,1 of the detectors mask; see the description of SDET command). The return value is unpredictable for any other value of item. Example GI 0 Byte sequence: Description: ISD Microcontroller 25 VoiceDSP 25 Get the duration of the last detected DTMF tone. The VoiceDSP processor responds: 60 ms. 00 AA AA 00 00 06 2-31 ISD-T360SB GL 2—SOFTWARE Get Length Returns the length of the current message in multiples of 4 Kbytes (blocks). The returned value includes the message directory information (64 bytes for the first block and 32bytes for every other block), the message data, and the entire last block of the message, even if the message occupies only a portion of the last block. Since a memory block includes 4096 bytes, the returned length may be bigger than the actual message length by up to 4095 bytes. The minimum length of a message is one block. Example GL Byte sequence: Description: Microcontroller VoiceDSP 19 AA AA 19 00 04 Get the length of the current message. The VoiceDSP processor responds: 4 i.e., the message occupies 16384 (4 * 4096) bytes. GMS Get Memory Status Returns the total remaining memory blocks as a 16bit unsigned integer. The estimated remaining recording time may be calculated as follows: Time = (Num_of_blocks x 4096 x 8) / ( Compression_rate x 1000) This estimate assumes no silence compression: a real recording may be longer, according to the amount of silence detected and compressed. Example GMS Byte sequence: Description: GMT Microcontroller 12 AA AA VoiceDSP Return the remaining memory blocks. The VoiceDSP responds: 40 blocks. 12 00 28 Get Message Tag Returns the 16-bit tag associated with the current message. If the current message is undefined, ERR_VALID is reported. 2-32 Voice Solutions in Silicon™ ISD-T360SB 2—SOFTWARE Example GMT Byte sequence: Description: Microcontroller VoiceDSP 04 AA AA 04 00 0E Get the current message tag. In a system where the message tag is encoded as described in the DMS command, the VoiceDSP processor return value indicates that the message is a new ICM in mailbox Number 6. Get Number of Messages tag_ref tag_mask GNM Returns the number of messages whose message tags match the tag_ref parameter. Only bits set in tag_mask are compared, a match is considered successful if: message tag and tag_mask = tag_ref and tag_mask where and is a bitwise AND operation. The tag_ref and tag_mask parameters are each two bytes; the return value is also two bytes long. See “Message Tag” on page 2-3 for a description of message-tag encoding. If tag_mask = 0, the total number of all existing messages is returned, regardless of the tag_ref value. Example GNM FFFE 0003 Byte sequence: Microcontroller 11 FF FE 00 03 AA AA 05 VoiceDSP Get the number of messages which have bit 0 cleared, and bit 1 set, in their message tags. VoiceDSP processor responds that there are five messages which satisfy the request. 11 Description: GSW FF FE 00 03 00 Get Status Word Returns the 2-byte status word. STATUS WORD The VoiceDSP processor has a 16-bit status word to indicate events that occur during normal operation. The VoiceDSP processor asserts the MWRQST signal (clears to 0), to indicate a change in the status word. This signal remains active until the VoiceDSP processor receives a GSW command. The status word is cleared during reset, and upon a successful GSW command. 15 14 13 12 11 10 9 8 7 6 5 4 EV_ DTMF EV_ RESET EV_ VOX EV_ CONST_ NRG Res EV_ MEMLOW EV_ DIALTONE EV_ BUSY EV_ ERROR EV_ MEMFULL EV_ NORMAL_ END EV_ DTMF_ END ISD 3 0 EV_ DTMF_ DIGIT 2-33 ISD-T360SB 2—SOFTWARE The bits in the status word are used as follows: EV_DTMF_DIGIT DTMF digit. A value indicating a detected DTMF digit. (See the description of DTMF code in the GT command.) EV_DTMF_END 1 = Ended detection of a DTMF tone. The detected digit is held in EV_DTMF_DIGIT. EV_NORMAL_END 1 = Normal completion of operation, e.g., end of message playback. EV_MEMFULL 1 = Memory is full. EV_ERROR 1 = Error detected in the last command. You must issue the GEW command to return the error code and clear the error condition. EV_BUSY 1 = Busy tone detected. Use this indicator for call progress and line disconnection. EV_DIALTONE 1 = Dial tone detected. Use this indicator for call progress and line disconnection. EV_MEMLOW 1 = Not enough memory. (See CMSG command for further details.) EV_CONST_NRG 1 = A period of constant energy was detected. Use this indicator for line disconnection. (See CONST_NRG_TIME_COUNT in Table 2-8.) EV_VOX 1 = A period of silence (no energy) was detected on the telephone line. Use this indicator for line disconnection. (See VOX_TIME_COUNT in Table 2-8.) EV_RESET When the VoiceDSP processor completes its power-up sequence and enters the RESET state, this bit is set to 1, and the MWRQST signal is activated (cleared to 0). Normally, this bit changes to 0 after performing the INIT command. If this bit is set during normal operation of the VoiceDSP processor, it indicates an internal VoiceDSP processor error. The microcontroller can recover from such an error by reinitializing the system. EV_DTMF 1 = Started detection of a DTMF tone. 2-34 Voice Solutions in Silicon™ ISD-T360SB 2—SOFTWARE Example GSW Byte sequence: Description: GT Microcontroller VoiceDSP 14 AA AA 14 00 40 Get the VoiceDSP processor Status Word (typically sent after the MMRQST signal is asserted by the VoiceDSP processor which indicates a change in the status word). The VoiceDSP processor responds that the memory is full. Generate Tone tone Generates the tone specified by the 1-byte tone parameter. The VoiceDSP state changes to TONE_GENERATE. The tone generation continues until an S command is received. A DTMF or a single frequency tone may be generated as shown: To generate a DTMF encode the bits as follows: Bit 0 1 Bits 1–4 DTMF code. Where the DTMF code is encoded as follows: Value (Hex)DTMF Digit 0 to 90 to 9 A A B * C # D B E C F D Bits 5–7 0 To generate a single frequency tone encode the bits as follows: Bit 0 0 Bits 1-5 3–30 The value in bits 1–5 is multiplied by 100 to generate the required frequency (300Hz–3000Hz). ISD 2-35 ISD-T360SB 2—SOFTWARE Bits 6-7 0 The VoiceDSP processor does not check for the validity of the tone specification. Invalid specification yields unpredictable results. Example GT 20 Byte sequence: Description: Microcontroller VoiceDSP 0D 20 0D 20 Generate a single-frequency 1600Hz tone. Get Time and Day time_day_option GTD Returns the time and day as a 2-byte value. time_day_option may be one of the following: 0 Get the system time and day. 1 Get the current message time-and-day stamp. Any other time_day_option returns the time-and-day stamp of the current message. Time and day are encoded as follows: Bits 0–2 Day of the week (1 through 7). Bits 3–7 Hour of the day (0 through 23). Bits 8–13 Minute of the hour (0 through 59). Bits 14–15 NOTE 2-36 00 The time was not set before the current message was recorded. 11 The time was set, i.e., the SETD (Set Time of Day) command was executed. If the current message is undefined, and time_day_option is 1, an ERR_INVALID error is reported. Voice Solutions in Silicon™ ISD-T360SB 2—SOFTWARE Example GTD 1 Byte sequence: Description: Microcontroller VoiceDSP 0E 01 AA AA 0E 01 E8 29 Get the current message time-and-day stamp. The VoiceDSP processor responds that the message was created on the first day of the week at 5:40 A.M. The return value also indicates that the SETD command was used to set the system time and day before the message was recorded. Note: If the SAS command is used to announce the time-and-day stamp, “Monday” is announced as the first day of the week. For an external vocabulary, the announcement depends on the vocabulary definition (See the IVS User’s Manual for more details). Get Tagged Message tag_ref tag_mask dir GTM Selects the current message, according to instructions in dir, to be the first, nth next or nth previous message which complies with the equation: message tag and tag_mask = tag_ref and tag_mask. where and is a bitwise AND operation. dir is one of the following: 0 Selects the first (oldest) message. -128 Selects the last (newest) message. n Selects the nth next message starting from the current message. –n Selects the nth previous message starting from the current message. To select the nth message with a given tag to be the current message you must first select the first message that complies with the above equation, and then issue another GTM command with n – 1 as a parameter, to skip to the nth message. NOTE To select the nth, or -nth, message with a given tag to be the current message you must first select the first message (dir=0), or the last message (dir=-128), that complies with the above equation, and then issue another GTM command with n-1 (for next message), or -n+1 (for previous message), as a parameter, to skip to the nth, or -nth, message respectively. \f a message is found, it becomes the current message and 1 (TRUE) is returned. If no message is found, the current message remains unchanged and 0 (FALSE) is returned. If dir is not 0, and the current message is undefined the return value is unpredictable. After the command execution the current message may either remain undefined or changed to any existing message. The only exception is when the GTM command is executed just after the DM command. (See the DM command for further details.) To access the nth message, when n > 127, a sequence of GTM commands is required. ISD 2-37 ISD-T360SB 2—SOFTWARE Example GTM FFCE 003F 0 Byte sequence: Description: 09 FF CE 00 3F 00 AA Microcontroller 09 FF CE 00 3F 00 01 VoiceDSP Select the oldest of the new ICMs, in mailbox number 6, to be the current message, for a system where the message tag is encoded as described in the example for the DMS command. The VoiceDSP processor returns a value indicates that there is such a message. The following pseudo-code demonstrates how to play all new ICMs in mailbox number 6. The messages are marked as old after being played: Return_val = GTM(FFCE, 003F, 00) /*Get the oldest message with the defined tag*/ While (ReturnVal == TRUE) Begin P() /* Play */ Message_tag = GMT() /* Get message tag */ SMT(FFF7) /* Mark the message as ‘old’ */ GTM(FFCE, 003F, 01) /* Get next message with the same tag */ End GTUNE Get Tune index Gets the value of the tunable parameter identified by index (one byte) as the 2-byte value, parameter_value. This command may be used to read and identify the parameter value that was set to tune the VoiceDSP. If index does not point to a valid tunable parameter, ERR_PARAM is set in the error word. The GTUNE command may be used in IDLE state or RESET state. If TUNE command was not used to set the tunable parameters, then the GTUNE command will read the default parameter value. Tables 2-4 to 2-11 describe the tunable parameters, their index numbers and their default values. Example GTUNE 17 Byte sequence: Description: 2-38 Microcontroller VoiceDSP Get the minimum period for busy detection ComactSPEECH responds: 700 (7 seconds). 06 17 AA AA 06 17 02 BC Voice Solutions in Silicon™ ISD-T360SB 2—SOFTWARE INIT Initialize System Execute this command after the VoiceDSP processor has been configured (see CFG command). INIT performs a soft reset of the VoiceDSP as follows: • Initializes the message directory information. • Messages are not deleted. To delete the messages, use the DM and DMS commands. • Sets the detectors mask to 0. • Sets the volume level that is controlled by the VC command, to 0. • Sets the playback speed to normal (0). • Switches to the IDLE state. • Initializes the tone detectors. The current message is undefined after INIT execution. The tunable parameters are not affected by this command. They are set to their default values only during RESET. Example INIT Byte sequence: Description: INJ Microcontroller VoiceDSP Initialize the VoiceDSP processor. 13 13 Inject IVS Data n byte1. . . byten Injects vocabulary data of size n bytes to good Flash blocks. This command programs Flash devices, on a production line, with IVS vocabulary data. It is optimized for speed; all VoiceDSP processor detectors are suspended during execution of the command. Use the CVOC command to check whether programming was successful. If there is not enough memory space for the vocabulary data, ERR_PARAM is set in the error word, and execution stops. Flash blocks that include IVS data can not be used for recording, even if only one byte of the block contains IVS data (e.g., if the vocabulary size is 4K + 100 bytes, two blocks of the Flash are not available for message recording). Example INJ 00000080 Data Byte sequence: Description: ISD 29 Microcontroller 29 VoiceDSP Inject 128 bytes of vocabulary data. 00 00 00 80 Vocabulary Data 00 00 00 80 Echo of Data 2-39 ISD-T360SB MR 2—SOFTWARE Memory Reset Erases all memory blocks and initializes the VoiceDSP processor (does exactly what the INIT command does). Bad blocks, and blocks which are used for IVS vocabularies, are not erased. This command can be issued in either RESET or IDLE states. NOTE When Memory Reset is used in RESET state, it must be issued after the CFG command is issued, or the memory type and number of devices are not defined. In this case the result is unpredictable. NOTE The command erases all messages and should be used with care. Example MR Byte sequence: Microcontroller 2A VoiceDSP Erase all memory blocks. 2A Description: P Playback Begins playback of the current message. The VoiceDSP processor state changes to PLAY. When playback is complete, the VoiceDSP processor sets the EV_NORMAL_END bit in the status word, and activates (clears to 0) the MWRQST signal. The state then changes to IDLE. Playback can be paused with the PA command, and can be resumed later with the RES command. Playback can be stopped with the S command. If the current message is undefined, ERR_INVALID is reported. Example P Byte sequence: Description: PA Microcontroller VoiceDSP Play the current message. 03 03 Pause Suspends the execution of the current GT, P, R, SAS, SO, SW, or SS command. The PA command does not change the state of the VoiceDSP processor; execution can be resumed with the RES command. NOTE 2-40 DTMF and tone detectors remain active during Pause. Voice Solutions in Silicon™ ISD-T360SB 2—SOFTWARE Example PA Byte sequence: Description: PDM Microcontroller VoiceDSP 1C 1C Suspend playback of current message. Go To Power-Down Mode Switches the VoiceDSP processor to power-down mode (see “POWER-DOWN MODE” on page 1-4 for details). Sending any command while in power-down mode resets the processor detectors, and returns it to normal operation mode. NOTE If an event report is pending (MWRQST is active) and not processed by the microcontroller prior to issuing the PDM command, the event is lost. Example PDM Byte sequence: Microcontroller 1A 1A Description: VoiceDSP Put the VoiceDSP processor in power-down mode. R Record tag compression_rate Records a new message with message tag tag and compression rate compression_rate. The VoiceDSP processor state changes to RECORD. The R command continues execution until stopped by the S command. Recording can be paused with the PA command, and can be resumed later with the RES command. If the memory becomes full, recording stops and EV_MEMFULL is set in the status word. See “Message Tag” on page 2-3 for a description of message-tag encoding. The compression rate may be defined as 0 for PCM recording or either 1, 3, 6 for compression rates 5.3 Kbits/sec, 9.9 Kbits/sec, 16.8 Kbits/sec respectively. See “ VCD” on page 2-10 for a description of the compression algorithm. NOTE A time-and-day stamp is automatically attached to each message. Before using the R command for the first time, use the SETD command. Failure to do so results in undefined values for the time-andday stamp. Example of a typical recording session: • (ICM) The microcontroller detects the first ring. • (ICM, OGM, memo) The microcontroller sends the R command. ISD 2-41 ISD-T360SB 2—SOFTWARE Example R 000E 03 Byte sequence: Description: RDET 0C 00 0E 03 Microcontroller 0C 00 0E 03 VoiceDSP Record a new ICM in mailbox Number 6 in a system where the message tag is encoded as described in the example of the DMS command. The compression rate is defined as 9.9 Kbits/sec. Reset Detectors detectors_reset_mask Resets the VoiceDSP processor tone and energy detectors according to the value of the detectors_reset_mask parameter. A bit set to 1 in the mask, resets the corresponding detector. A bit cleared to 0 is ignored. The 1-byte detectors_reset_mask is encoded as follows: Bit 0 Reset the busy and dial tone detectors. Bits 1–3 Reserved. Must be cleared to 0. Bit 4 Reset the constant energy detector. Bit 5 Reset the no energy (VOX) detector. Bit 6 Reset the DTMF detector. Bit 7 Reserved. Must be cleared to 0. Example RDET 20 Byte sequence: Description: 2-42 Microcontroller VoiceDSP Reset the VOX detector. 2C 20 2C 20 Voice Solutions in Silicon™ ISD-T360SB 2—SOFTWARE RES Resume Resumes the activity that was suspended by the PA, SB, or SF commands. Example RES Byte sequence: Description: RMSG Microcontroller VoiceDSP Resume playback which was suspended by either the PA, SF or SB command. 1D 1D Read Message Returns 32 bytes of data from the current position of the message pointer, and advances the message pointer by 32 bytes. If the VoiceDSP processor was in the IDLE state, the command opens the current message, switches the VoiceDSP processor to the MSG_OPEN state, sets the message pointer to the beginning of the message data, and returns the 32 bytes of data. The microcontroller must issue an S command to close the message, and switch the VoiceDSP processor to the IDLE state. If the current message is undefined, ERR_INVALID is reported. Trying to read beyond the end of the message sets the EV_NORMAL_END event, and the VoiceDSP processor switches to the IDLE state. In this case, the return value is undefined and should be ignored. Example RMSG Byte sequence: Description: S AA AA Microcontroller 32 VoiceDSP 32 32 bytes of data Read 32 bytes from the current message memory. ... Stop Stops execution of the current command and switches the VoiceDSP processor to the IDLE state. S may be used to stop the execution of CMSG, SMSG, WMSG, RMSG and all asynchronous commands. Example S Byte sequence: Description: ISD 00 Microcontroller 00 VoiceDSP Stop current activity (e.g., playback, recording) and change the VoiceDSP processor to IDLE state. 2-43 ISD-T360SB SAS 2—SOFTWARE Say Argumented Sentence sentence_n arg Announces sentence number sentence_n of the currently selected vocabulary, and passes arg. sentence_n and arg are each 1-byte long. The VoiceDSP processor state changes to SYNTHESIS. When playing is complete, the VoiceDSP processor sets the EV_NORMAL_END bit in the status word, and activates the MWRQST signal. The state then changes to IDLE. If the current vocabulary is undefined, ERR_INVALID is reported. Example SAS 00 03 Byte sequence: Microcontroller Description: 1E 00 03 VoiceDSP Announce the first sentence in the sentence table of the currently selected vocabulary with ‘3’ as the actual parameter. SB 1E 00 03 Skip Backward time_length Skips backward in the current message time_length units, in 0.2 second segments, and pauses message playback. The RES command must be issue to continue playback. time_length is a 2-byte parameter that can have any value up to 320 (64 seconds). The skip accuracy is five percent. SB is meaningful only in the PLAY state. If the beginning of the message is detected during the execution of the SB command, execution terminates, the EV_NORMAL_END bit in the status register sets, the MWRQST signal activates, and the processor switches to the IDLE state. If time_length is greater than 320, ERR_PARAM is set in the error word. Playback speed does not affect the behavior of this command. Example SB 0019 Byte sequence: Microcontroller Description: 23 00 19 VoiceDSP Skip backwards five seconds from the current position in the message being played. SDET 23 00 19 Set Detectors Mask detectors_mask Controls the reporting of detection of tones and energy detectors according to the value of the detectors_mask parameter. A bit set to 1 in the mask, enables the reporting of the corresponding detector. A bit cleared to 0 disables the reporting. Disabling reporting of a detector does not stop or reset the detector. The 1-byte detectors_mask is encoded as follows: 2-44 Voice Solutions in Silicon™ ISD-T360SB 2—SOFTWARE Bit 0 Report detection of a busy tone. Bit 1 Report detection of a dial tone. Bit 2-3 Reserved. Must be cleared to 0. Bit 4 Report detection of a constant energy. Bit 5 Report detection of no energy (VOX) on the line. Bit 6 Report the ending of a detected DTMF. Bit 7 Report the start of a detected DTMF (up to 40 ms after detection start). Example SDET A3 10 Byte sequence: Microcontroller 10 VoiceDSP Description: Set reporting of all VoiceDSP processor detectors, except for end-of-DTMF. SE B3 B3 Skip to End of Message This command is valid only in the PLAY state. When invoked, playback is suspended (as for the PA command), and a jump to the end of the message is performed. Playback remains suspended after the jump. Example SE Byte sequence: Description: ISD Microcontroller VoiceDSP 24 24 Skip to end of current message. 2-45 ISD-T360SB SETD 2—SOFTWARE Set Time and Day time_and_day Sets the system time and day as specified by the 2-bytes time_and_day parameter. The time_and_day parameter is encoded as follows: Bits 0–2 Day of the week (1 through 7). Bits 3–7 Hour of the day (0 through 23). Bits 8–13 Minute of the hour (0 through 59). Bits 14–15 Must be set to 1. If time_and_day value is not valid, ERR_PARAM is set in the error word. Example SETD DE09 Byte sequence: Description: SF 0F DE 09 Microcontroller 0F DE 09 VoiceDSP Set time and day to Monday 1.30 A.M. (where Monday is the first day of the week) Skip Forward time_length Skips forward in the current message time_length units, in 0.2 second segments, and causes message playback to pause. The RES command must be issue to continue playback. time_length is a 2-byte parameter that can have any value up to 320 (64 seconds). The skip accuracy is five percent. This command is meaningful only in the PLAY state. The RES command must be issue to continue playback. If the end of the message is detected during execution of SF, execution of the command is terminated the EV_NORMAL_END bit in the status word is set, the MWRQST signal is activated and the processor switches to the IDLE state. If time_length is greater than 320, ERR_PARAM is set in the error word. Playback speed does not affect the behavior of this command. Example SF 0019 Byte sequence: Description: 2-46 22 00 19 Microcontroller 22 00 19 VoiceDSP Skip forward five seconds from the current position in the message being played. Voice Solutions in Silicon™ ISD-T360SB 2—SOFTWARE Set Message Pointer num_of_pages SMSG Sets the message pointer to num_of_pages x 32 bytes from the beginning of the current message data. If the VoiceDSP processor was in the IDLE state, the command opens the current message and switches the VoiceDSP processor to the MSG_OPEN state. The microcontroller must issue an S command to close the message, and switch the VoiceDSP processor to the IDLE state. If num_of_pages x 32 is greater than the message length, EV_NORMAL_END is set in the status word, the message pointer is set to the end of the message, and the VoiceDSP processor switches to the IDLE state. If the current message is undefined, ERR_INVALID is reported. Example SMSG 000A Byte sequence: Description: Microcontroller 30 00 0A VoiceDSP 30 00 0A Set the message pointer to 10 pages (320 bytes) from the beginning of the current message data. Set Message Tag message_tag SMT Sets the tag of the current message. The 2-byte message_tag can be used to implement mailbox functions by including the mailbox number in the tag, or to handle old and new messages differently by using one bit in the tag to mark the message as old or new. See “Message Tag” on page 2-3. To change the message tag, you should first get the tag using the GMT command, read the tag, modify it, and write it back. NOTE Message tag bits can only be cleared. Message tag bits are set only when a message is first created. If the current message is undefined, ERR_INVALID is reported. Example SMT FFF7 Byte sequence: Description: 05 FF F7 Microcontroller 05 FF F7 VoiceDSP Mark the current message as old in a system where the message tag is encoded as described in the example of the DMS command. Note that the VoiceDSP processor ignores bits in the tag which are set to 1; only bit 3 is modified in the message tag. ISD 2-47 ISD-T360SB 2—SOFTWARE Say One Word word_number SO Plays the word number word_number in the current vocabulary. The 1-byte word_number may be any value from 0 through the index of the last word in the vocabulary. The VoiceDSP processor state changes to SYNTHESIS. When playback of the selected word has been completed, the VoiceDSP processor sets the EV_NORMAL_END bit in the status word, and activates the MWRQST signal. The state then changes to IDLE. If word_number is not defined in the current vocabulary, or if it is an IVS control or option code, ERR_PARAM is set in the error word. If the current vocabulary is undefined, ERR_INVALID is reported. Example SO 00 Byte sequence: Microcontroller Description: VoiceDSP Announce the first word in the word table of the currently selected vocabulary. 07 07 00 00 Set Playback Speed speed SPS Sets the speed of message playback as specified by speed. The new speed applies to all recorded messages and synthesized messages (only if synthesized using IVS), until changed by another SPS command. If this command is issued while the VoiceDSP processor is in the PLAY state, the speed also changes for the message currently being played. Speed may be one of 13 values, from –6 to +6. A value of 0 represents normal speed. NOTE A negative speed value represents an increase in speed, a positive value represents a decrease in speed. NOTE The playback speed control is not applicable when the stored messages or the IVS data are not compressed (Stored in PCM format). The change in speed is approximate, and depends on the recorded data. In any case, if i < j, playback speed with parameter i is the same or faster than with parameter j. If speed is not in the –6 to +6 range, ERR_PARAM is set in the error word. Example SPS FB Byte sequence: Description: 2-48 Microcontroller VoiceDSP Set playback speed to –5. 16 FB 16 FB Voice Solutions in Silicon™ ISD-T360SB 2—SOFTWARE Say Sentence sentence_n SS Say sentence number sentence_n of the currently selected vocabulary. sentence_n is 1-byte long. The VoiceDSP processor state changes to SYNTHESIS. If the sentence has an argument, 0 is passed as the value for this argument. When playing has been completed, the VoiceDSP processor sets the EV_NORMAL_END bit in the status word, and activates the MWRQST signal. The state then changes to IDLE. If sentence_n is not defined in the current vocabulary, ERR_PARAM is set in the error word. If the current vocabulary is undefined, ERR_INVALID is reported. Example SS 00 Byte sequence: Description: Microcontroller 1F 00 VoiceDSP 1F 00 Announce the first sentence in the sentence table of the currently selected vocabulary. Set Speakerphone Mode mode SSM Sets the speakerphone to the mode mode of operation. The command is valid when the VoiceDSP processor is in IDLE state. mode can be one of: 0 OFF Deactivate the speakerphone, and return the VoiceDSP processor to normal operation mode. 1 ON Put the VoiceDSP processor in speakerphone mode and activate speakerphone in full-duplex mode i.e., with full cancellation of both the acoustic and the electrical echoes. Tone detectors are not active. Gains in the Send and Receive paths are set by the relevant tunable parameters. 2 TRANSPARENT Activate the speakerphone with no echo cancellation (this mode is used for system tuning). 3 MUTE Activate the speakerphone, while generating silence to the line. The near-end-listener can hear the far-end-speaker, but not vice versa. Tone detectors are not active. 4 LISTEN The line is audible on the speaker. Tone detectors are active. This mode is used for call generation. 5 Reserved. 6 RESTART Restart the current speakerphone mode. This mode differs from ON; it does not require full initialization of the speakerphone. It should be used to resume the speakerphone operation after HOLD mode or to adjust to an environment change (e.g., parallel pickup). 7 HOLD Stop the codec interrupts. Neither side can hear each other. See “Full-duplex Speakerphone” on page 2-14 for more details. ISD 2-49 ISD-T360SB NOTE 2—SOFTWARE Only commands that are specified in Table 2-3, are active during all speakerphone modes (other than 0). Example SSM 01 Byte sequence: Microcontroller 2F 01 VoiceDSP 2F 01 Put the VoiceDSP processor into Speakerphone mode, and set the speakerphone to full-duplex mode. Description: Set Vocabulary Type type id SV Selects the vocabulary table to be used for voice synthesis. The vocabulary type is set according to the 1-byte type parameter: 0 For compatibility only. 1 External vocabulary in ROM. 2 External vocabulary in Flash. 3-7 Reserved. The host is responsible for selecting the current vocabulary, with SV command, before using an SAS, SO, SS or SW command. Each external vocabulary table has a unique id which is part of the vocabulary internal header (See the IVS User’s Guide for more details). If type is 1 or 2, the VoiceDSP processor searches for the one byte id parameter in each vocabulary table header until a match is found. If the id parameter does not point to a valid IVS vocabulary, ERR_PARAM is set in the error word. Example SV 02 03 Byte sequence: Description: SW 20 02 03 Microcontroller 20 02 03 VoiceDSP Select the vocabulary with vocabulary-id 3, which resides on a Flash, as the current vocabulary. Say Words n word1. . . wordn Plays n words, indexed by word1 to wordn. The VoiceDSP processor state changes to SYNTHESIS. On completion, the EV_NORMAL_END bit in the status word is set, and the MWRQST signal goes low. The state then changes to IDLE. If one of the words is not defined in the current vocabulary, or if it is an IVS control or option code, or if n > 8, ERR_PARAM is reported. 2-50 Voice Solutions in Silicon™ ISD-T360SB 2—SOFTWARE If the current vocabulary is undefined, ERR_INVALID is reported. Example SW 02 00 00 Byte sequence: Description: TUNE 21 02 00 00 Microcontroller 21 02 00 00 VoiceDSP Announce the first word, in the word table of the currently selected vocabulary, twice. Tune index parameter_value Sets the value of the tunable parameter identified by index (one byte) to the 2-byte value, parameter_value. This command may be used to tune the DSP algorithms to a specific Data Access Arrangement (DAA) interface, or to change other parameters. If you do not use TUNE, the VoiceDSP processor uses default values. If index does not point to a valid tunable parameter, ERR_PARAM is set in the error word. NOTE The tunable parameters are assigned with their default values on application of power. The INIT command does not affect these parameters. The following tables 2-4 to 2-11 describe the tunable parameters, their index numbers and their default values, grouped by their functionality. Table 2-4: TUNABLE PARAMETERS: Voice Compression and Decompression (VCD) Index Parameter Name Description Default 4 Voice Activity Detection (VAD): VAD_SIL_THRESHOLD Prevents speech from being interpreted as silence. The silence detection algorithm has an adaptive threshold, which is changed according to the noise level. This parameter is, therefore, only the initial threshold level. Legal values: 9216 to 13824 in 512 (6 dB) steps. 11264 5 Voice Activity Detection (VAD): VAD_SIL_THRESHOLD_ST EP Defines the adaptive threshold changes step. If this threshold is too low, the threshold converges too slowly. If it is too high, silence detection is too sensitive to any noise. Legal values: 3 to 48. 12 6 Voice Activity Detection (VAD): VAD_SIL_BURST_THRESH OLD The minimum time period for speech detection, during silence. As this threshold increases, the time period interpreted as silence increases. If this threshold is too low, a burst of noise is detected as speech. If it is too high, words may be partially cut off. Legal values: 1 to 3. 2 ISD 2-51 ISD-T360SB 2—SOFTWARE Table 2-4: TUNABLE PARAMETERS: Voice Compression and Decompression (VCD) Index Parameter Name Description Default 7 Voice Activity Detection (VAD): VAD_SIL_HANG_THRESH OLD The minimum time period for silence detection, during speech. As this threshold increases, the time period interpreted as silence decreases. If this threshold is too low, words may be partially cut off. If it is too high, no silence is detected. Legal values: 8 to 31. 15 8 Voice Activity Detection (VAD): VAD_SIL_ENABLE Silence compression control. 0 turns silence compression off. Note: Silence compression must be turned off when using ARAM for voice storage. Otherwise the playback quality is unpredictable. 1 9 Voice Activity Detection (VAD): VAD_ENERGY_FACTOR Determines the energy level used to synthesize silence. For the default value, the energy levels of the synthesized silence and the recorded silence are the same. If you divide (multiply) the default value by two, the synthesized silence is 6 dB less (more) than the level of the recorded silence. Legal values: 1024 to 16384. 8192 70 SW Automatic Gain Control (SW AGC): SWAGC_ENABLE SW AGC control. 0 turns SW AGC off. 1 11 SW Automatic Gain Control (SW AGC): SWAGC_FACTOR Determines the maximum gain that the SW AGC algorithm may use. Legal values: 0, 1, 2, 4, 8, 16, 32, 64, 128 Note: Value 0 means the the maximum gain is defined by the algorithm. 128 2-52 Voice Solutions in Silicon™ ISD-T360SB 2—SOFTWARE Table 2-5: TUNABLE PARAMETERS: Tone Generation and Message Playback Index Parameter Name Description Default 27 DTMF Generation: A one-byte value that controls the twist level of a DTMF tone, 66 DTMF_GEN_TWIST_LEVEL generated by the GT command, by controlling the energy level of each of the two tones (low frequency and high frequency) composing the DTMF tone. The Least Significant Nibble (LSN) controls the low tone and the Most Significant Nibble (MSN) controls the high tone. The energy level of each tone, as measured at the output of a TP3054 codec (before the DAA) connected to the VoiceDSP processor is summarized in the following table: Nibble Value Tone Energy (dB-Volts) 0 0 1 –17.8 2 –14.3 3 –12.9 4 –12.4 5 –12.0 6 –11.9 7 –11.85 8..15 –11.85 The volume of the generated DTMF tone during measurements was 6. (TONE_GEN_LEVEL+VOL_LEVEL = 6). For the default level, the high tone is –14.3 dBV and the low tone is –12.4 dBV, which gives a DTMF twist level of 1.9 dB. The energy level of a single generated tone is the level of the low tone. 16 Tone Generation: TONE_GEN_LEVEL Controls the energy level at which DTMF and other tones are generated. Each unit represents 3 dB. The default level is the reference level. For example, if you set this parameter to 4, the energy level is 6 dB less than the default level. The actual output level is the sum of TONE_GEN_LEVEL and the VOL_LEVEL variable, controlled by the VC command. The tones are distorted when the level is set too high. Legal values: 0 ≤ TONE_GEN_LEVEL + VOL_LEVEL ≤ 12. 6 21 VCD Playback and Voice Synthesis: VCD_PLAY_LEVEL Controls the energy during playback and external voice synthesis. Each unit represents 3 dB. The default level is the reference level. For example, if you set this parameter to 4, the energy level is 6 dB less than the default level. The actual output level is the sum of VCD_LEVEL and the VOL_LEVEL variable, controlled by the VC command. Speech is distorted when the level is set too high. Legal values: 0 ≤ VCD_PLAY_LEVEL + VOL_LEVEL ≤ 12. 6 ISD 2-53 ISD-T360SB 2—SOFTWARE Table 2-6: TUNABLE PARAMETERS: DTMF Detection Index Parameter Name Description Default 17 Energy Level: DTMF_DET_MIN_ENERG Y Minimum energy level at which DTMF tones are detected. If you divide (multiply) the value by 2, the detection sensitivity decreases (increases) by 3 dB. Legal values: 8 to 4096 32 24 Echo Canceler: DTMF_DET_ECHO_DELA Y The near-echo delay in samples. The sampling rate is 8000Hz (i.e., 125 ms per sample). Legal values: 0 to 16. 4 26 Twist Level: DTMF_DET_REV_TWIST Controls the reverse twist level at which the VoiceDSP processor detects DTMF tones. While the normal twist is set at 8 dB, the reverse twist can be either 4 dB (default) or 8 dB (if this parameter is set to 1). 0 60 SW AGC: DTMF_DET_AGC_IDLE SW AGC for DTMF in idle/record modes. When incrementing the tunable by 1, the dynamic range is increased by 3 dB. Legal values: 0 to 5. 0 61 SW AGC: DTMF_DET_AGC_PLAY Software AGC for play mode and tone generation modes. When incrementing the tunable by 1, the dynamic range increases by 3 dB. Legal values: 0 to 16. 3 2-54 Voice Solutions in Silicon™ ISD-T360SB 2—SOFTWARE Table 2-7: TUNABLE PARAMETERS: Tone Detection Index Parameter Name Description Default 18 Dial Tone: Controls the duration of a tone before it is reported as a dial 700 TONE_DET_TIME_COUNT tone, in 10 msec units. The accuracy of the constant is ±10 ms. Legal values: 0 to 65535. 19 Busy and Dial Tone: TONE_DET_ON_ENERGY _ THRESHOLD Minimum energy level at which busy and dial tones are detected as ON (after 700Hz filtering). If you divide (multiply) the value by 2 you get about 3 dB decrease (increase) in the threshold. The mapping between energy level and the parameter value is as follows (measured on the codec output when a 400Hz tone was injected to the codec input): Tunable value Energy threshold (dB-Volts) 10 –31.8 20 –28.6 100 –21.7 500 –14.7 8000 –2.5 Legal values: 0 to 65535. 160 20 Busy and Dial Tone: TONE_DET_OFF_ENERGY _ THRESHOLD Maximum energy level at which busy and dial tones are detected as OFF (after 700Hz filtering). If you divide (multiply) the value by 2 you get about 3 dB decrease (increase) in the threshold. The mapping between energy level and the parameter value is the same as for TONE_ON_ENERGY_THRESHOLD Legal values: 0 to 65535. 110 23 Busy Tone: BUSY_DET_MIN_TIME Minimum time period for busy detection, in 10 ms units. The accuracy of the constant is ±10 ms. Legal values: 0 to 65535. 600 53 Busy Tone: BUSY_DET_MIN_ON_TIM E Minimum period considered as On period for busy tone detection. Note that for weak signals: (–30 dB and below) the maximum value is 12 (i.e., 120 ms minimum detection time). Unit: 10 ms. Accuracy is ±20 ms. Legal values: 10 to 1000. 10 54 Busy Tone: Maximum period considered as On for busy-tone detection. BUSY_DET_MAX_ON_TIM Unit: 10 ms. Accuracy is ±20 ms. E Legal values: 10 to 1000. ISD 168 2-55 ISD-T360SB 2—SOFTWARE Table 2-7: TUNABLE PARAMETERS: Tone Detection Index Parameter Name Description Default 55 Busy Tone: Minimum period considered as Off for busy-tone detection. BUSY_DET_MIN_OFF_TIM Unit: 10 ms. Accuracy is ±20 ms. E Legal values: 5 to 1000. 7 56 Busy Tone: BUSY_DET_MAX_OFF_TI ME Maximum period considered as On for busy-tone detection. Unit: 10 ms. Accuracy is ±20 ms. Legal values: 5 to 1000. 122 57 Busy Tone: BUSY_DET_VERIFY_COU NT Number of On/Off cadences that must be detected prior to reporting busy-tone presence. Legal values: 9 to 127. 9 58 Busy Tone: BUSY_DET_TONE_TYPE Specifies the type of busy tone to detect: 1 —Two cadences 2 —Three cadences 3 —Both two and three cadences 1 59 Busy Tone: BUSY_DET_DIFF_THRESH OLD The maximum allowed difference between two compared On or Off periods. Unit: 10 ms. Legal values: 0 to 1000. 9 2-56 Voice Solutions in Silicon™ ISD-T360SB 2—SOFTWARE Table 2-8: TUNABLE PARAMETERS: Energy Detection Index Parameter Name Description Default 10 Silence (VOX): This parameter determines the minimum energy level at VOX_DET_ENERGY_THRE which voice is detected. Below this level, it is interpreted as SHOLD silence. Legal values: 1 to 32767. 12 12 Silence (VOX): VOX_DET_TIME_COUNT 700 22 Silence (VOX): Controls the maximum energy-period, in 10 ms units, that does VOX_DET_TOLERANCE_T NOT reset the vox detector. IME Legal values: 0 to 255. 3 47 Constant Energy: Minimum elapsed time until the VoiceDSP processor reports CONST_NRG_DET_TIME_ constant energy level. Units: 10 ms. Accuracy: ±10 ms Legal values: 1 to 65534 COUNT 700 48 Constant Energy: CONST_NRG_DET_ TOLERANCE_TIME Variations in constant energy, up to this time, do not reset the constant energy detector. Units: 10 ms. Legal values: 0 to 255 5 49 Constant Energy: CONST_NRG_DET_LOW _ THRESHOLD Determines the minimum energy level that is treated as 1 constant energy. The minimum energy is calculated as follows: (1—1/2CONST_NRG_DET_LOW_THRESHOLD) * average_energy Legal values: 1 to 16 50 Constant Energy: CONST_NRG_DET_HIGH _ THRESHOLD Determines the maximum energy level that is treated as constant energy. The maximum energy is calculated as follows: (1 + 1/2CONST_NRG_DET_HIGH_THRESHOLD ) * average_energy Legal values: 0 to 16 ISD This parameter, in units of 10 ms, determines the period of silence before the VoiceDSP processor reports silence. The accuracy of the constant is ±10 ms. Legal values: 0 to 65535. 1 2-57 ISD-T360SB 2—SOFTWARE Table 2-9: TUNABLE PARAMETERS: Speakerphone Index Parameter Name Description Default 31 Acoustic Echo Canceler (AEC): SP_AEC_PRIORITY_BIAS Controls the bias in priority between the Send and Receive paths. If send priority-bias is preferred, the value should be greater than zero. For no priority bias, the value should be zero. For priority bias for the Receive path, the value should be negative. Steps are 3 dB each (e.g., +3 is 9 dB bias for the Send path, –2 is 6 dB bias for the Receive path). Legal values: –4 to 4. 0 32 Acoustic Echo Canceler (AEC): SP_AEC_COUPLING_ LOSS_THRESHOLD This parameter limits the acoustic return loss. Its value (SP_AEC_COUPLING_LOSS_THRESHOLD / 32767) is compared with the RMS value of Sout divided by the RMS value of Rin, during a single-talk event. The loop gain is decreased, if necessary, to control the TCL level. For SP_AEC_COUPLING_LOSS_THRESHOLD = 32767 this loop is disabled. Legal values: 0 to 32767. 2047 34 Acoustic Echo Canceler (AEC): SP_AEC_LR_LEVEL Controls the speakerphone gain from the microphone to the line-out. The total attenuation, or gain, depends on both of the analog gains and this value. The gain is: K * signal where: K = SP_AEC_LR_LEVEL/4096. Legal values: 0 to 16000. 14000 36 Acoustic Echo Canceler (AEC): SP_AEC_CLIP_POS Specifies the positive peak-value at which the analog circuit of the line-out saturates. Codec analog full scale corresponds to µLAW full scale values after expansion.Assume that positive saturation occurs at amplitudes higher than those of a sine wave at X [dBm0]. The SP_AEC_CLIP_POS value is set as: SP_AEC_CLIP_POS = 32636 * 10((X – 3.17)/20) Note: a sine wave with amplitude 4 * 8159 = 32636 corresponds to 3.17 dBm0. Example: For X = –6.2761 dBm0, the value is: SP_AEC_CLIP_POS = 32636 * 10((–6.2761– 3.17)/20) = 0.3371 * 32636 =11000; Legal values: 0 to 32767. 16000 37 Acoustic Echo Canceler (AEC): SP_AEC_CLIP_NEG Specifies the negative peak value at which the analog circuit of the line-out saturates. Codec analog full scale corresponds to µLAW full scale values after expansion.The value of SP_AEC_CLIP_NEG is set as shown for SP_AEC_CLIP_POS, above. Legal values: –32768 to 0. –16000 2-58 Voice Solutions in Silicon™ ISD-T360SB 2—SOFTWARE Table 2-9: TUNABLE PARAMETERS: Speakerphone Index Parameter Name Description Default 40 Acoustic Echo Canceler (AEC): SP_AEC_ENABLE Enables/disables the acoustic echo controller. Legal values: 0 (disable), 1 (enable). 1 43 Acoustic Echo Canceler (AEC): SP_AEC_VOX_HYST Controls the hysteresis in near-talker detection. (The speakerphone state machine has a built-in hysteresis mechanism to prevent fluctuations in the talker identification process i.e., identifying the active side.) The value of this parameter is a dimensionless number, which should be evaluated during the tuning process for specific hardware. Larger values for the parameter correspond to a wider hysteresis loop. Negative values increase the probability that the state machine remains in the last state. Legal values: –127 to 127. 10 45 Acoustic Echo Canceler (AEC): SP_AEC_DTD_TH Controls the sensitivity of the system. Low values correspond to 73 high sensitivity, with a greater false alarm probability (i.e., an echo is considered a real talker). High values correspond to low sensitivity, with slower switching. This parameter is affected by the loop gain and the specific hardware characteristics. Legal values: 0 to 127. 35 Electric Echo Canceler (EEC): SP_EEC_LR_LEVEL Controls the speakerphone gain from the line-in to the speaker. The total attenuation, or gain, depends on both of the analog gains and this value. The gain is: K * signal where: K = (SP_EEC_LR_LEVEL/4096) * (2(6 + VOL_LEVEL)/2) Legal values: 0 to 400. ISD 281 2-59 ISD-T360SB 2—SOFTWARE Table 2-9: TUNABLE PARAMETERS: Speakerphone Index Parameter Name Description Default 38 Electric Echo Canceler (EEC): SP_EEC_CLIP_POS Specifies the positive peak value at which the analog circuit of the speaker saturates. Codec analog full scale corresponds to µLAW full scale values after expansion. The value of SP_EEC_CLIP_POS is set as shown for SP_AEC_CLIP_POS, above. Legal values: 0 to 32767. 16000 39 Electric Echo Canceler (EEC): SP_EEC_CLIP_NEG Specifies the negative peak value at which the analog circuit of the line-out saturates. Codec analog full scale corresponds to µLAW full scale values after expansion. The value of SP_EEC_CLIP_NEG is set as shown for SP_AEC_CLIP_POS, above. Legal values: –32768 to 0. –16000 41 Electric Echo Canceler (EEC): SP_EEC_ENABLE Enables/disables the electrical echo controller. Legal values: 0 (disable), 1 (enable). 1 44 Electric Echo Canceler (EEC): SP_EEC_VOX_HYST Controls the hysteresis in far-talker detection. (The speakerphone state machine has a built-in hysteresis mechanism to prevent fluctuations in the talker identification process i.e., identifying the active side.) The value of this parameter is a dimensionless number, which should be evaluated during the tuning process for specific hardware. Larger values for the parameter correspond to a wider hysteresis loop. Negative values increase the probability that the state machine remains in the last state. Legal values: –127 to 127. 10 46 Electric Echo Canceler (EEC): SP_EEC_DTD_TH Controls the sensitivity of the system. Low values correspond to 82 high sensitivity, with a greater false alarm probability (i.e., an echo is considered a real talker). High values correspond to low sensitivity, with slower switching. This parameter is affected by the loop gain and the specific hardware characteristics. Legal values: 0 to 127. 33 Attenuation: SP_BLOCK_LEVEL Controls the maximum attenuation level of the speakerphone suppressors. It affects the speakerphone stability and its subjective quality. The maximum attenuation is calculated according to: SP_BLOCK_LEVEL/228 Legal values: 550 to 32000. 10922 42 Tone Generation: SP_TONE_GEN_ LEVEL Controls the energy level at which DTMF, and other tones, are generated to the line (codec 0) while the speakerphone is active. Each unit represents 3 dB. Legal values: 0 ≤ SP_TONE_GEN_LEVEL ≤ 10. Note: the energy level at which the tones are generated to the speaker (codec 1) while the speakerphone is active, is controlled by the TONE_GEN_LEVEL tunable parameter and the vol_level. 6 2-60 Voice Solutions in Silicon™ ISD-T360SB 2—SOFTWARE Table 2-10: TUNABLE PARAMETERS: Memory Support Index 62 Parameter Name Description Memory Device Size: Defines the nubber of blocks (each block is of 4096 bytes) in NUM_OF_BLOCKS_IN_M every memory device (Flash or ARAM/DRAM). The number EM and type of connected devices are defined by the CFG command. Default 128 Flash Device Size (Mbits) Number of Blocks Value 4 128 8 256 16 512 ARAM/DRAM Device Size (Mbits) Number of Blocks Value 16 508 63 Memory Size for Testing Defines the nubber of blocks (each block is of 4096 bytes) in NUM_OF_BLOCKS_FOR_ every memory device (Flash or ARAM/DRAM) for production TEST line testing purposes. The number should be small to minimize testing time during the production sequence. However, the number of blocks should be larger than the number of expected bad blocks in the memory device. In case of value=0, no productiontest is performed. In any case other than value= 0, the number of blocks is defined by the parameter value, and a production testing cycle is performed after RESET. Legal values: 0 to 128. Note: If power fails during production testing cycle, the memory status is unpredicted. The memory device should be replaced and the production test should be repeated. 64 ARAM Quality Level: MAX_DEFECT_NIBBLES_I N_BLOCK ISD 0 Defines the maximum allowed bad nibbles in ARAM block 0 (each block is of 8192 nibbles). A nibble (4 bits) is considered bad if any bit is defected. If the number of bad nibbles in a block exceeds the maximum allowed value, the block is marked as bad block and is not used for voice storage. Legal values: 0 to 255. 2-61 ISD-T360SB 2—SOFTWARE Table 2-11: TUNABLE PARAMETERS: Codec Support (Samples) Index Parameter Name Description Default 65 Channel 0 Delay: CFRD0 The delay of codec channel 0 from Frame Synch 0 (CFS0) to start of valid data. Legal values: 0 to 255 1 66 Channel 1 Delay: CFRD1 The delay of codec channel 1 from Frame Synch 0 (CFS0) to start of valid data. Legal values: 0 to 255 10 67 Channel 2 Delay: CFRD2 The delay of codec channel 2 from Frame Synch 0 (CFS0) to start of valid data. Legal values: 0 to 255 10 68 Frame Synch Delay: CFSD The delay of Frame Synch 1 (CFS1) from Frame Synch 0 (CFS0). 10 Legal values: 0 to 255 69 Data Valiid Delay: CFET The delay between Frame Synch 0 (CFS0) to end of valid data 18 of all channels. Legal values: 0 to 255 Example TUNE 17 02BC Byte sequence: Description: VC 15 17 Microcontroller 15 17 VoiceDSP Set the minimum period for busy detection to 700 (7 seconds). 02 BC 02 BC Volume Control vol_level Controls the energy level of all the output generators (playback, tone generation, and voice synthesis), with one command. The resolution is ±3 dB. The actual output level is composed of the tunable level variable, plus the vol_level. The valid range for the actual output level of each output generator is defined in Table 2-5. For example, if the tunable variable VCD_LEVEL (parameter number 21) is 6, and vol_level is –2, then the output level equals VCD_LEVEL + vol_level = 4. Example VC 04 Byte sequence: Description: 2-62 Microcontroller VoiceDSP Set the volume level to VCD_LEVEL + 4. 28 04 28 04 Voice Solutions in Silicon™ ISD-T360SB 2—SOFTWARE Write Message data WMSG Writes 32 bytes of data to the current position of the message pointer, and advances the message pointer by 32 bytes. If the VoiceDSP processor is in the IDLE state, the command opens the current message, switches the VoiceDSP processor to the MSG_OPEN state, sets the message pointer to the beginning of the message data, and writes the 32 bytes of data. To add data at the end of an existing message, issue the SMSG command to the last page of the message. Issue the WMSG command with a buffer consisting of 32 FF bytes (this has no effect on the current data in the page). A subsequent WMSG command adds a new block to the message, and writing continues at the beginning of the new block. The microcontroller must issue an S command to close the message and switch the VoiceDSP processor to the IDLE state. NOTE When updating an existing message, bits can only be cleared, but not set. If the current message is undefined, ERR_INVALID is reported. Example WMSG 32 bytes Byte sequence: Microcontroller Description: VoiceDSP Write 32 bytes in the message memory. 31 31 ISD 32 bytes of data to write echo 32 bytes of data 2-63 ISD-T360SB 2-64 2—SOFTWARE Voice Solutions in Silicon™ 3—SCHEMATIC DIAGRAMS ISD-T360SB Chapter 3ÑSCHEMATIC DIAGRAMS 3.1 APPLICATION INFORMATION The following schematic diagrams for a VoiceDSP processor Reference design unit. This reference design includes three basic clusters: • An 80C51 MicroController. • VoiceDSP processor cluster, including a TP3054 codec, and an ISDT360SB controlling a Flash device. • User interface that includes one 16-digit LCD, and a 16-key (4 x 4) keypad. ISD 3-1 ISD-T360SB 3-2 3—SCHEMATIC DIAGRAMS Voice Solutions in Silicon™ 4—PHYSICAL DIMENSIONS ISD-T360SB Chapter 4ÑPHYSICAL DIMENSIONS Figure 4-1: 80-Pin Plastic Quad Flat Package, Top and Bottom—Type: Metric PQFP, 14x14 Body 1. ISD All dimensions are in millimeters. All dimensions and tolerances conform to ANSI Y14.5-1982. 4-1 ISD-T360SB 4—PHYSICAL DIMENSIONS Figure 4-2: 80-Pin Plastic Quad Flat Package, Side—Type: Metric PQFP, 14x14 Body Table 4-1: Packaging Dimensions Symbol Min. Nom. Max. A — 2.82 3.00 A1 0.10 0.15 0.25 A2 2.55 2.67 2.75 D 17.20 BSC. D1 14.00 BSC. D2 12.35BSC. ZD 0.825 REF. E 17.20 BSC. E1 14.00 BSC. E2 12.35 BSC. ZE 0.825 REF. L 0.88 N 80 e 0.65 BSC. b 0.22 b1 0.22 ccc 4-2 0.73 1.03 0.38 0.30 0.33 0.12 Voice Solutions in Silicon™ IMPORTANT NOTICES The warranty for each product of ISD (Information Storage Devices, Inc.), is contained in a written warranty which governs sale and use of such product. Such warranty is contained in the printed terms and conditions under which such product is sold, or in a separate written warranty supplied with the product. Please refer to such written warranty with respect to its applicability to certain applications of such product. These Product may be subject to restrictions on use. Please contact ISD, for a list of the current additional restrictions on these Product. By purchasing these Product, the purchaser of these Product agrees to comply with such use restrictions. Please contact ISD for clarification of any restrictions described herein. ISD, reserves the right, without further notice, to change the ISD ChipCorder product specifications and/or information in this document and to improve reliability, functions and design. ISD assumes no responsibility or liability for any use of the ISD ChipCorder Product. ISD conveys no license or title, either expressed or implied, under any patent, copyright, or mask work right to the ISD ChipCorder Product, and ISD makes no warranties or representations that the ISD ChipCorder Product are free from patent, copyright, or mask work right infringement, unless otherwise specified. The 100-year retention and 100K record cycle projections are based upon accelerated reliability tests, as published in the ISD Reliability Report, and are neither warranted nor guaranteed by ISD. Information contained in this ISD ChipCorder data sheet supersedes all data for the ISD ChipCorder Product published by ISD prior to December, 1998. This data sheet and any future addendum to this data sheet is (are) the complete and controlling ISD ChipCorder product specifications. In the event any inconsistencies exist between the information in this and other product documentation, or in the event that other product documentation contains information in addition to the information in this, the information contained herein supersedes and governs such other information in its entirety. Copyright© 1998, ISD (Information Storage Devices, Inc.) All rights reserved. ISD is a registered trademark of ISD. ChipCorder is a trademark of ISD. All other trademarks are properties of their respective owners. Application examples and alternative uses of any integrated circuit contained in this publication are for illustration purposes only and ISD makes no representation or warranty that such applications shall be suitable for the use specified. 2045 Hamilton Ave. San Jose, California 95125-5904 Tel: 408/369-2400 Fax: 408/369-2422 http://www.isd.com Part No. 2201298D5008