Electronic Design and Prototyping Services Back to Prototypes

We offer our customers a unique competitive advantage in our ability to program and apply all of the major industrial/consumer microcontrollers from Taiwan. The following paragraphs highlight the features of the controllers available.

SONIX's SN68000 wave file player chips with 4-bit microcontroller and a dual-tone music and can store from 3 to 340 sec of sounds. Their SN66000 series 4-channel wave file player and wave table synthesis chip is available in 12 to 340 sec storage time. SN69000/6A000 wave file player with direct LCD drive can store from 12 to 340 sec of sounds and drive from 128 to 1024 LCD dots (segments). SONIX also has the excellent SN65000 series, storing 4 to 20 sec of sounds with direct 8-ohm speaker drive. All the SONIX sound IC's are based a 4-bit microcontroller so they can be programmed to solve complex user interaction and control problems.

Winbond's POWER SPEECH wave file player chips can do simple on/off control and can store from 3 to 60 sec of speech or sounds (@ 6 kHz sample rate) Their BAND DIRECTOR 4-channel wave file player and wave table synthesis chips have are programmable for more complex behavior and can store from 15 sec to 2 min of sounds. Their VIEW TALK series can play wave files has a dual-tone synthesizer and is capable of direct LCD drive. Sound storage is from 15 to 60 sec, LCD segment drive from 768 to 2048. Winbond has the W58300 series of direct-speaker-drive wave file speech ICs with 12 to 128 sec of sound storage. Winbond has an impressive line of 4-bit microcontrollers in low and standard operating voltage versions and 80C51 versions for low and normal voltage operation.

Elan Microelectronics Corporation (EMC) has ICs in many categories usable in consumer applications. Their EM57000 series wave file player with 4-bit microcontroller also has dual-tone music and a unique sound-effects generator. Their EM58000 series is a 2-channel wave file player with dual-tone. These two IC series are available in 3 to 340 sec sound storage times. The EM22000 series is a 6-60 sec direct speaker drive IC. EMC also has available a long list of 4- and 8-bit microcontroller chips with voice and direct LCD drive.

Hualon Electronics Ltd (HEC) has an excellent line of 8-bit consumer-oriented microcontrollers with direct LCD drive and 1- and 2-channel 8-bit Digital-to-Analog Converters (DACs) for speech synthesis.

SUNPLUS has several varieties of its dual-DAC MOS Technology/Rockwell International 6502 look-alike controllers useful for any application in which you require speech, voice-piano music, and/or direct LCD drive.

These products are all impressive performers. Reliability and quality are excellent.

Note that wave file speech is distinguished from Linear Predictive Coding (LPC) speech and the global standard for this vocal-tract-emulation technology, CELP from Codebook Excitation Linear Predictive coding. The LPC and CELP chips are more expensive for any product with less than 2-3 minutes of stored speech. Wave file players store digital samples of the sound in on-board Read Only Memory (ROM). To play sounds, the chip retrieves the samples from memory and outputs them through its Digital to Analog Converter (DAC). This is how your computer sound card plays wave files. So if the stored sound has been sampled at 6000 samples per second and if the samples are ADPCM-encoded then you must use 24,000 bits of storage for each second of speech stored. Adaptive Differential Pulse Code Modulation (ADPCM) is a redundancy-removing speech compression method in which the absolute value of the sound waveform sample is not stored, only the difference between the current sample and nearby samples is stored. LPC and CELP remove redundancy even more by emulating the human vocal tract and speech production mechanism. With LPC good voice quality can be achieved at data rates as low as 2400 bits per second. The speech may sound somewhat robotic and, of course music cannot be stored but singing can. Also, the stored speech requires somewhat time-consuming and thus expensive preparation. The CELP algorithm is in the public domain, requires only simple sound processing to store, and can reproduce music. With CELP excellent results can be achieved at, say 6700 bits per sec.

Dual-tone music is synthesized music created by a simple square-wave oscillator. It's computer-generated music. The dual-tone in the name comes from the fact that 2 tones can be produced at the same time, For best dual-tone sound generator should output a stepped approximation to a sinewave. If a square-wave is output certain harmonies in the music will sound somewhat sour. EMC, SONIX, and Winbond all have good dual-tone generators.

Wavetable synthesis music is, generally, realistic sounding music. In this method two or more instrument waveforms are stored for each instrument, a low note and a high note.. The stored notes are partitioned into an attack portion which is played once when the note is struck, a middle or sustain portion that is looped to hold the note, and a release portion (a version of the sustain portion) that is played to fade the note out. This is the way the music synthesizer in your computer works when it plays midi (Musical Instrument Digital Interface) files. (Unless it's using the outdated but less expensive fm synthesizer technique.)

The so-called "voice-piano" method of speech storage works by storing 1 or more wavefiles of an instrument sound. This sound file is then just played back at different rates a slower rate to play a low note, a higher rate to play the sound as a high note. This is the technique generally used whenever you hear a barking dog or mewing cat singing a song in a product. As the sound is pitched down it gets longer. As it is pitched up, it gets shorter. Your computer sound card probably uses this technique to play percussion (drum) sounds which don't have an easily-identifiable and loopable sustain portion.

Also note that if dual-tone or wavetable synthesis music is stored in the chip long playing times can be achieved since only the note pitch and its duration must be stored. For wavetable or voice-piano synthesis the basic sounds must of course be stored.

Back to Prototypes