[sci.electronics] Question about speech Synthesisor

bkc@sun.soe.clarkson.edu (Brad Clements) (04/14/88)

I just returned from my local Radio Shack store, where I was looking
over their speech synthesisor and controller chip. I didn't bye the
set because they require a 3.12 Mhz and 10 Mhz crystal, both of
which would have to be special ordered and might take over a month
to get.

What I'd like to know is if anyone has experience with these chips.
How well does it sound and how well is the text converted to speech.
Has anyone incorporated a 'exception EPROM' with their own special
words to alonomes (?) conversions etc?

Lastly, does anyone know where I can find these chips or similar
construction materials (kits even) for less or sooner?

Any information would be appreciated, I am, after all, a basement
experimentor and not well versed in the microprocesser field.

knt@cbterra.ATT.COM (Kirk Trost) (04/14/88)

Brad Clements had asked some questions about the Radio Shack speech
synthesizer and text-to-speech converter.

First of all, the speech synthesizer does not require that a 3.12 Mhz
crystal be used -- as long as the crystal is pretty close to that value,
the synthesizer should work the same, only that the pitch of the 
resulting voice will be different.  I have used the speech synthesizer
with the commonly available (I believe Radio Shack still sells them)
3.58 Mhz colorburst crystal.  The pitch of the voice is higher than with
the 3.12 Mhz, but the quality is about the same.

Speaking of quality, the output of the speech synthesizer is not bad
but does lack somewhat is understandability.  If you know what the chip
is supposed to say, you can understand it pretty well, but to a person
that does not know what it is trying to say, a bit of confusion often 
results.

Nevertheless, the chip is extremely fun to play with and interfaces rather
easily to a microprocessor.  I know Brad said he is not well versed in
microprocessors, but I feel this is the best way to use the chip.  
Originally, I used just some static RAM and an address counter to make the
chip talk, but this method did not have the versatility that can be
obtained through using a microprocessor.

I have not used the text-to-speech converter and thus can not comment
on how well it works.  I have looked over the specifications on using the
converter in Radio Shack's Semiconductor Reference Manual, but don't
have the application to use it at this time.

Have fun and good luck!

**************************************************************************
Kirk Trost	AT & T  Bell Laboratories, Columbus, Ohio
cbosgd!knt
**************************************************************************

tedk@ihuxv.ATT.COM (Kekatos) (04/15/88)

In article <732@sun.soe.clarkson.edu> bkc@sun.soe.clarkson.edu (Brad Clements) writes:
|I just returned from my local Radio Shack store, where I was looking
|over their speech synthesisor and controller chip. I didn't bye the
|set because they require a 3.12 Mhz and 10 Mhz crystal, both of
|which would have to be special ordered and might take over a month
|to get.

There was a series of articles here on the net about 2 months ago. Maybe
someone out here saved a copy of the articles. Sorry, I didn't.

There is a board available (built, not a kit) for the IBMPC. The entire thing
is only about $70. It's called COMPU-TALKER. It uses the same chip
set. It can be used with any computer by adding your own power
supply. Call B.G. MICRO at (214) 271-5546 for more info.

|What I'd like to know is if anyone has experience with these chips.
|How well does it sound and how well is the text converted to speech.
|Has anyone incorporated a 'exception EPROM' with their own special
|words to alonomes (?) conversions etc?

I own the COMPU-TALKER, I feel that I sounds fine, for the cost.

|Lastly, does anyone know where I can find these chips or similar
|construction materials (kits even) for less or sooner?

There are several books published about all the chips. One is titled
"Chip Talk".

|Any information would be appreciated, I am, after all, a basement
|experimentor and not well versed in the microprocesser field.

I am assuming that you own a computer. You will need something to 
send data to your Speech Synthesis box.  There have got to be
other boards and/or kits available.


Ted G. Kekatos
backbone!ihnp4!ihuxv!tedk                     (312) 979-0804
AT&T Bell Laboratories, Indian Hill South, IX-1F-460
Naperville & Wheaton Roads - Naperville, Illinois. 60566 USA

flaig@cit-vlsi.Caltech.Edu (Charles M. Flaig) (04/15/88)

In article <732@sun.soe.clarkson.edu> bkc@sun.soe.clarkson.edu (Brad Clements) writes:
>I just returned from my local Radio Shack store, where I was looking
>over their speech synthesisor and controller chip. I didn't bye the
>set because they require a 3.12 Mhz and 10 Mhz crystal, both of
>which would have to be special ordered and might take over a month
>to get.
>
>What I'd like to know is if anyone has experience with these chips.
>How well does it sound and how well is the text converted to speech.
>Has anyone incorporated a 'exception EPROM' with their own special
>words to alonomes (?) conversions etc?

I used these two chips to build an RS-232 speech synthesizer for a school
project a few years back, and was reasoanably pleased with the results.

I recall using a 3.58MHz color-burst crystal rather than the 3.12MHz one
specified, and it worked fine.  And while I used the recommended 10MHz
crystal, I believe it just sets the speed on the controller chip and 
should work at a nearby frequency such as 8MHz (not guaranteed, just a
guess).

When it worked, it worked well.  Creative spelling often allowed it to
pronounce almost any word correctly.  Unfortunately it badly slurred r's
as I recall (ie, "three" came out as "thlwee") but this is so obviously
wrong that I'm sure I damaged the chip slightly during handling (since
I never bother with any precautions).

I intended to do more with it, such as programming my own EPROM, but
somehow just never got around to it....

______________________________________________________________________________
  ___   ,               ,                                           ,;,;;;,
 /   Y /|              /|              Charles Flaig                ;/@-@\;
|      |/     __, ,__  |/              flaig@csvax.caltech.edu       | ^ |
|      /^\   /  | | |  /  /\  /\                                      \=/ 
 \____/|  \_/|_/\_/ \_/ \_\/_/_/_/     "What, you think they PAY me for this?"

f12012ag@deimos.unm.edu (Chang H. Park) (04/15/88)

In article <732@sun.soe.clarkson.edu> bkc@sun.soe.clarkson.edu (Brad Clements) writes:
>I just returned from my local Radio Shack store, where I was looking
>over their speech synthesisor and controller chip. I didn't bye the
>set because they require a 3.12 Mhz and 10 Mhz crystal, both of
>which would have to be special ordered and might take over a month
>to get.
            The Radio Shack "TV Colorburst Crystal" works at 3.579545 MHz.
I used one in my setup...it worked fine.  From the ads I have seen, both
chips are on sale...well worth the price!  They work very well.  It is
also fun to substitute different value crystals in place of the 3.12 MHz        
crystal.  A little hunting at local electronics shops should produce the
10 MHz.   Good luck!

Ollie Eisman - N6LTJ

--

>What I'd like to know is if anyone has experience with these chips.
>How well does it sound and how well is the text converted to speech.
>Has anyone incorporated a 'exception EPROM' with their own special
>words to alonomes (?) conversions etc?
>
>Lastly, does anyone know where I can find these chips or similar
>construction materials (kits even) for less or sooner?
>
>Any information would be appreciated, I am, after all, a basement
>experimentor and not well versed in the microprocesser field.


SEDS-UNM : Students for the Exploration and Development of Space           
Box 92 Student Union, University of New Mexico, Albuquerque, NM  87106
(505) 277-3171

wte@ncrcae.Columbia.NCR.COM (Bill Eason) (04/16/88)

In article <3862@cbterra.ATT.COM> knt@cbterra.UUCP (Kirk Trost) writes:
>
>Brad Clements had asked some questions about the Radio Shack speech
>synthesizer and text-to-speech converter.
>
>First of all, the speech synthesizer does not require that a 3.12 Mhz
>crystal be used -- as long as the crystal is pretty close to that value,
>the synthesizer should work the same, only that the pitch of the 
>resulting voice will be different...

Sometime last fall I bought just the speech chip (sans text converter)
and started building a simple circuit to play with it until I get a PC
to control it.  I was a little disappointed with the application notes
supplied by RS with the chip.  They hinted that the chip had a good bit
more capability than they told how to use.  If anyone has more app-note
information or knows where to get it, please post or Email and I will
summarize.  More specifically:  (Bear with memory lapses; it's been a 
few months since I've looked at the thing...)

1)  Is there any way to control the pitch of the voice other than modifying
	crystal frequency?  I remember the specs talking about the chip
	being able to emulate a vocal tract, but would that imply the
	ability to modulate pitch, too?

2)  What is the eighth data (or did they call that address) bit for?  If
	memory serves me correctly, there were only 64 or 128 phonemes(?)
	allophones(?) available in the specs, which leaves a line or two
	of additional data which could be fed to the chip.  Anyone know
	what these are for and how to use them?

3)  The specs gave a rough diagram of connection to external ROM, but
	didn't say much more than "It can be done."  That may be enough
	to impress some people, but my inquiring mind wants to know HOW!!
	What kind of data format goes into these ROMs?  Pulse-width data?
	At what sampling rate?  What kind of block marker/header/size-limit
	is there to delineate one phoneme or phrase from the next?  It is
	at this level that I get frustrated with the computer-store-with-
	a-few-discrete-parts-in-the-back shop that RS is (or has become).

>... I have looked over the specifications on using the
>converter in Radio Shack's Semiconductor Reference Manual, but don't
>have the application to use it at this time.

Maybe I should check here, but I somehow doubt that it would give me a whole
lot more than the specs/"app notes" that came with the chip.  Ted G. Kakatos
posted a followup saying that a book, _Chip_Talk_, goes into more detail.  
Any more info on what it tells, where to find it?

"Anything you can do to help would be . . . helpful"  -Holy Grail

TIA,  Bill
-- 

Bill Eason   (803) 791-6917    ...!ucbvax!sdcsvax!ncr-sd!ncrcae!util5!wte
NCR Corporation		  ...!rutgers!mcnc!ece-csc!ncrcae!util5!wte
E & M Columbia   3325 Platt Springs Rd.   West Columbia, SC  29169

flaig@cit-vlsi.Caltech.Edu (Charles M. Flaig) (04/18/88)

In article <3265@ncrcae.Columbia.NCR.COM> wte@ncrcae.UUCP (Bill Eason) writes:

[concerning the RS speech synthesizer chip set]

>3)  The specs gave a rough diagram of connection to external ROM, but
>	didn't say much more than "It can be done."  That may be enough
>	to impress some people, but my inquiring mind wants to know HOW!!
>	What kind of data format goes into these ROMs?  Pulse-width data?
>	At what sampling rate?  What kind of block marker/header/size-limit
>	is there to delineate one phoneme or phrase from the next?  It is
>	at this level that I get frustrated with the computer-store-with-
>	a-few-discrete-parts-in-the-back shop that RS is (or has become).

As I recall, there was an comment at the end of the included application
notes listing an address where you could request the additional application
notes necessary for using external ROMs.  I would suspect that the ROMs
contain rule tables (such as the examples given in the application notes)
which translate word fragments (or whole words) into the allonemes (?sp)
required by the speech chip.

______________________________________________________________________________
  ___   ,               ,                                           ,;,;;;,
 /   Y /|              /|              Charles Flaig                ;/@-@\;
|      |/     __, ,__  |/              flaig@csvax.caltech.edu       | ^ |
|      /^\   /  | | |  /  /\  /\                                      \=/ 
 \____/|  \_/|_/\_/ \_/ \_\/_/_/_/     "What, you think they PAY me for this?"

john@anasaz.UUCP (John Moore) (04/26/88)

For all who are wondering about the radio shack chip,
it is produced by the folks who used to be General Instrument Semiconductors
(and have recently done an LBO and changed the name to Microchip Technologies).
Check with them (Chandler, Arizona - 602-963-7373) or a rep. The chip is an
SP0256-AL2 and they do have specs available (although the writing is
not very good). I have used this chip in one product and am not very
happy with it - the speech quality is poor making it hard to distinguish
certain words (ONE vs DONE, for example). However, it is widely available
and easy to interface. I have not used the text-to-speech micro that they
sell as a companion to it.
-- 
John Moore (NJ7E)   hao!noao!mcdsun!nud!anasaz!john
(602) 870-3330 (day or evening)
The opinions expressed here are obviously not mine, so they must be
someone else's.