Reference STATE OF THE ART Stan Augarten

ISBN 0-89919-195-9

Photo of
Another Way to Record Sound
A Digital Speech Synthesizer, the S3610 LPC-10


Just as ICs can store and process numbers and words, so they can store and play back sounds - whether speech or music or audio effects like the noise of an exploding spaceship in a video game. Such sound-storing and playback chips are known as digital synthesizers, and they have been around since 1978, when Texas Instruments introduced the first single-chip speech synthesizer. TI's creation was first used in a company product called "Speak and Spell" - a toy that caused quite a stir and became very popular.

  The principle behind sound synthesis is known as linear predictive coding (the LPC in the name of the chip shown here). Any sound, whether speech, music, or audio effects, is a kind of information, with unique attributes that can be characterized digitally; in the case of speech, those attributes include the speaker's inflection, pronunciation, and pacing. To store speech in a digital synthesis system, a tape recording of the desired words is encoded by the chip manufacturer into binary pulses; these, in turn, are stored in a ROM. Some speech synthesizer chips contain their own ROM, others use separate memory ICs.

  To replay the words so that they can be heard, the synthesizer chip must be inserted in an electronic system that, at the very least, contains a loudspeaker, a power source, a capacitor, a resonator, and a microprocessor. When the system is turned on, the synthesizer chip releases its contents and, with the help of the other components, causes the loudspeaker to vibrate in a way that recreates the original sounds.

  In theory, digital synthesizers can achieve near-perfect levels of fidelity, more or less forever. Digital systems have no moving parts - no records to warp, no phonograph needles to wear out - so they don't suffer from the usual operational vagaries of phonographs and tape recorders. They are also much smaller, contain fewer parts, and are relatively inexpensive. At the moment, most digital synthesizers are unable to produce high-quality sounds and lack the storage capacity of conventional recording media. They are constantly being improved, however, and eventually will probably replace all other forms of sound storage and reproduction equipment. Since engineers have also developed speech recognition chips, computers that can talk, and listen, are on the horizon.

The S3610 speech synthesizer chip, shown here on a wafer, can store up to thirty-two words, or seventeen seconds' worth of speech, in its own 20K ROM (the large grid on the right). Hooked up to four separate 128K ROMs, its storage capacity can be boosted to seven minutes. Most of the circuits on the chip are used to decode information stored in the ROM and play it back. Actual size: 0.170 x 0.279 inches. Photo of


©Copyright Stan Augarten
This book is provided for general reference. The National Museum of American History and the Smithsonian Institution make no claims as to the accuracy or completeness of this work.

page:   Index    2   4   6   8   10   12   14   16   18   20   22   24   26   28   30   32   34   36   38   40   42   44   46   48   50   52   54   56   58   60   62   64   66   68   70   72   74   76  

National Museum of American History

HomeSearchChip TalkChip ArtPatentsPeoplePicturesCreditsCopyrightComments