SSI 263 phoneme speech synthesizer


na podstawie dokumentacji SSI 263 z SILICON SYSTEMS - innovators in integration:


DESCRIPTION

The SSI 263 is a versatile, high-quality, phoneme-based speech synthesizer circuit contained in a single monolithic CMOS integrated circuit. It is designed to produce an audio output of unlimited vocabulary, music and sound effects at an extremely low data input rate.

Speech is synthesized by combining phonemes, the building blocks of speech, in an appropriate sequence. The SSI 263 contains five eight-bit registers that allow software control of speech rate, pitch, pitch movement rate, amplitude, articulation rate, vocal tract filter response, and phoneme selection and duration.

FEATURES


wiecej informacji technicznych na ewentualne zyczenie.


oprocz tego jeszcze krotka notka jak go praktycznie wykorzystywac:


The Atari 1400XL/1450XLD Internal Voice Synthesizer.

Information complied on 4/8/90 by:
John Hardie

*NOTE* Some of this information was gathered from the Aug. 83 issue of Compute!

Being an avid collector of Atari Vapor-ware, it gave me great pleasure to finally get my 1450XLD up and going. Now the only problem was learning how to access the internal modem and voice synthesizer. After searching all by back issues of Antic and Analog with no luck, a friend turned me on to the Aug. 83 issue of Compute!. Compute!???, I said with a look of disbelief, but sure enough on page 33 I found some incredible information. Let me convey this info to you, my fellow vapor-warist.

Up until this point, I knew you could open the V: device for input with a simple basic command like this:
10 OPEN #1,8,0,"V:"

This command would OPEN a CHANNEL (1) for OUTPUT (8) to the Voice Chip (V:). The V: device can only be opened for output. Then to make it say something you would print to that channel like this:
10 OPEN #1,8,0,"V:"
20 ? #1,CHR$(42);CHR$(2);CHR$(0);CHR$(31);CHR$(42)

The above example will say the word "TEST". As this is obviously a time-consuming task, I quickly discovered that you could put the actual ATASCII symbols into a string in it's place, ie.
20 ? #1;"*5&)*"

This does not say anything intelligible, but is used just as an example. After reading the Compute! article and doing some follow-up research on my own, I discovered that there is more to this little baby than meets the eye or ear! The example in Compute! said to open the channel like this:
OPEN #1,8,0,"V1:PF"

"V1:PF"??? What the hell is that? I was convinced that someone at Compute! was pulling my leg and checked to make sure I wasn't really reading the April issue. The article mentions 3 speech modes. Supposedly, opening a channel with "V1:PF" opens the channel in Phoneme mode. Try this example:
10 OPEN #1,8,0,"V1:PF"
20 ? #1;"UHTAHREE"

The synthesizer should say "ATARI". In this mode, the computer ignores certain consonants which might confuse the synthesizer, such as "C" and "X". For a soft "C", you must use an "S"; for a hard "C", a "K". Similarly, an "X" is spelled "EKS". I assume that the "P" in the above example stands for PHONEME, but I don't know what the "F" stands for. The article also mentions 2 other modes: ALPHA and NUMERIC. ALPHA is a more direct text-to-speech mode. and NUMERIC allows voice programming in machine language. The article doesn't say how to access these other modes, though. Here's where the experimenting came in to play.

I found that V1: is the same as V: just as D1: is the same as D:. The "F" in "V1:RF" doesn't seem to have any effect. You could open the channel with "V:P" for the same results. It might be an optional parameter of some sort, although I have discovered that you can use "D" and "S" instead of "F". You can access the ALPHA mode by opening "V:S". I assume the "S" stands for "SPEECH". In this mode, you can: ? #1,"ATARI" and you should hear "ATARI". Even in this mode, you won't get perfect translation but using phonemes where necessary should be a lot easier than printing a bunch of character strings as in the earlier example. The NUMERIC mode is achieved by: OPEN #1,8,0,"V:N". This is the same mode that you would get if you just opened "V:". I assume that this is the default mode. As I showed at the beginning of this text you would have to print a lot of character strings to access the voice this way. The following chart lists the different modes and syntax:

Remember - V: and V1: are the same
NUMERIC mode - OPEN #1,8,0,"V:N" or OPEN #1,8,0,"V:"
PHONEME mode - OPEN #1,8,0,"V:P"
SPEECH mode - OPEN #1,8,0,"V:S"

Also remember there is an optional "F", "D", or "S" after these statements. Example: OPEN #1,8,0,"V:(N,P or S)(D,F or S)". I will try to research the seemingly optional parameter further and report in an addenum. I also found another mode although I don't konow it's function. "U". You can do a: OPEN #1,8,0,"V:U" but the output seems to be the same as "V:N" of just "V:". I don't know what "U" stands for (Useless, Unknown, Un-documented?), but again I will try to report in an addendum to this text.

A word listing might be found at a Heath/Zenth store since the Hero-1 robot from them uses the same Votrax SC-01 voice chip.

John Hardie

krAp, Thu Dec 26 16:57:22 MET 1996