Maker Pro
Maker Pro

Looking for CTS256 (text to speech processor from the 80-ies)

G

GdB

The CTS256-AL2 is a text to code processor. It converts English text
to phoneme codes used by the
SP0256-AL2, which is a speech generator. It is all stuff from the 80-
ies, but it would be great fun to do a little project with this
chip... Any speech chips appreciated. Please let me know if you can
help me.

Thanks.

George
 
S

Sjouke Burry

GdB said:
The CTS256-AL2 is a text to code processor. It converts English text
to phoneme codes used by the
SP0256-AL2, which is a speech generator. It is all stuff from the 80-
ies, but it would be great fun to do a little project with this
chip... Any speech chips appreciated. Please let me know if you can
help me.

Thanks.

George
Maybe you can find something on the
Texas Instruments TMS5220 with a custom Acorn ROM
I played with in the eighties/nineties in a bbc-b computer.
That unbelievable computer had a socket reserved just
for that purpose.
And no, I dont have the chip or computer anymore.....
 
R

Rafael Deliano

Any speech chips appreciated.

Best results for me: lurking at
ebay.com Electronic Components and/or Vintage Computing
with keywords like "Speech" and/or partnumber.
Only in very tough cases like Digitalker scanning the
www with google got better results.

Currently available at ebay:
General Instruments SP1000 ( LPC-lattice Analyzer, Syntesizer )
Interstate ASA16 ( filterbank for recognition )
General Instruments SP0250 ( LPC-lattice Synth. )
General Instruments SP0256AL2 ( LPC-lattice Synth. with Phonem. )
Votrax SC01A ( Formant Synth. with Phonem. )
I can scan datasheets and other documentation for
most of these oldtimers.
CTS256-AL2 , SP0256-AL2
Have been at ebay in the last year, but expensive.
And even if they had been cheaper: the SP0256-AL2
with its PWM-output had always a lousy quality compared
to Votrax.
The CTS256 probably implements the Naval Research Algorithm [1]
on a PIC7041 microcontroller.
For anyone using the SP0256-AL2 directly there has been an
Application Note containing precoded a list of common words [2].

MfG JRD

[1] Elovitz et Al. "Automatic Translation of English Text to Phonetics
by mean of letter to sound rules" United States Naval Research
Laboratory Report 7948 year: 1976
( don´ t have it, but would be interestred )
[2] Janet May "Allophone Speech Syntesis Technique" General Instruments
1982
( can scan it )
 
G

GdB

Currently available at ebay:
Saw these as well, all quite expensive, it's just for fun and I'm sure
someone has them in a drawer...
   General Instruments SP1000 ( LPC-lattice Analyzer, Syntesizer )
   Interstate ASA16 ( filterbank for recognition )
   General Instruments SP0250 ( LPC-lattice Synth. )
   General Instruments SP0256AL2 ( LPC-lattice Synth. with Phonem. )
   Votrax SC01A ( Formant Synth. with Phonem. )
I can scan datasheets and other documentation for
most of these oldtimers.
Wow, that would be useful! I got the SC01 as well as its datasheet.
CTS256-AL2 , SP0256-AL2

Have been at ebay in the last year, but expensive.
And even if they had been cheaper: the SP0256-AL2
with its PWM-output had always a lousy quality compared
to Votrax.
The CTS256 probably implements the Naval Research Algorithm [1]
on a PIC7041 microcontroller.
A PIC??? Or doesn't this have a relation with today's PIC's...
For anyone using the SP0256-AL2 directly there has been an
Application Note containing precoded a list of common words [2].
I've got the whole datasheet. If someone needs it. It holds the whole
list of allophones and examples of words to allophones...

MfG JRD

[1] Elovitz et Al. "Automatic Translation of English Text to Phonetics
     by mean of letter to sound rules" United States Naval Research
     Laboratory Report 7948  year: 1976
     ( don´ t have it, but would be interestred )
[2] Janet May "Allophone Speech Syntesis Technique" General Instruments
     1982
     ( can scan it )

Regards,

George
 
R

Rafael Deliano

Currently available at ebay:
Saw these as well, all quite expensive,
The SP1000 from Voyager are cheap.
For the others one can contact the seller
to see if one can find something agreeable
for both sides. Note that they and ebay
aren´t in it for hobby and apart from the
Votrax there isn´t that much sold.
that would be useful!
Datasheet:
http://www.embeddedFORTH.de/temp/sp1000-datasheet.pdf
Application Note:
http://www.embeddedFORTH.de/temp/sp1000-an.pdf
Manual:
http://www.embeddedFORTH.de/temp/sp1000-manual-1.pdf
http://www.embeddedFORTH.de/temp/sp1000-manual-2.pdf
http://www.embeddedFORTH.de/temp/sp1000-manual-3.pdf
My board:
http://www.embeddedFORTH.de/temp/sp1000p.pdf
My unfinished description (in german):
http://www.embeddedFORTH.de/temp/sp1000.pdf
Avoid the PWM-filter in the article, doesn´t work.

The problem with the SP1000 is that i and some other people
found it not working. But there was an article about it by
Steve Ciarcia in Byte and there seem to have been some
applications that used it. Its unclear if there are
batches of faulty chips or if the datasheet is wrong.
Even if it would work: the LPC-synthesizer has the usual
problem with the lousy PWM output. The gradient
lattice LPC analyzer is highly suspect too.
Datasheet:
http://www.embeddedFORTH.de/temp/asa-16.pdf
Article:
http://www.embeddedFORTH.de/temp/asa16-assp.pdf
My board:
http://www.embeddedFORTH.de/temp/asa16p.pdf
My description (in german):
http://www.embeddedFORTH.de/temp/asaapp.pdf
Found it easy to use. But make shure power-up
sequencing for its 3 supplies is properly done.
Datasheet should be easy to find in www.
There is a good description of it in the book
Bristow "Electronic Speech Synthesis"
My board:
http://www.embeddedFORTH.de/temp/sc01ap.pdf
My description ( in german ):
http://www.embeddedFORTH.de/temp/votrax.pdf

Other:
VCP200 zero crossing speech recognition IC
speaker independent, fixed vocabulary
There is a datasheet, but haven´s scanned it yet
My board:
http://www.embeddedFORTH.de/temp/vcp200p.pdf
My description ( in german ):
http://www.embeddedFORTH.de/temp/vcp.pdf
This is the lowest cost approach and one should not
expect too much.

HM2007 4 channel filterbank speech recognition IC
speaker dependent / trainable
Datasheet should be easy to find in www.
My board:
http://www.embeddedFORTH.de/temp/hm2007p.pdf
My description ( in german ):
http://www.embeddedFORTH.de/temp/hm2007p.pdf
The datasheet and the IC are slightly buggy.

Digitalker Mozer speech synthesizer
There are several datasheets for the Digitalker,
too many to list them here.
My board Digitalker:
http://www.embeddedFORTH.de/temp/digitalkerp.pdf
My board UAA1003:
http://www.embeddedFORTH.de/temp/uaa1003p.pdf
My description ( in german ):
http://www.embeddedFORTH.de/temp/mozer.pdf
There was another earlier Mozer IC from Telesensory.
The UAA1003 is a talking clock available in english and
german versions.

I have included pictures of my boards so one see what
external circuits are necessary. But note: aliasing filters
are mostly not included.
The layouts could be a good start for anyone using Eagle 4.15.
The CTS256 probably implements the Naval Research Algorithm [1]
on a PIC7041 microcontroller.
A PIC??? Or doesn't this have a relation with today's PIC's...

Anything there is called a PIC even a MIPS.
Texas Instruments tried in those days to introduce their 8 bit
TMS70xx controllers with much fanfare and General Instruments was
their second source. TI dropped them soon afterwards and so
hardly anyone remembers them.
For anyone using the SP0256-AL2 directly there has been an
Application Note containing precoded a list of common words [2].
I've got the whole datasheet. If someone needs it. It holds the whole
list of allophones and examples of words to allophones...
Probably these:
http://www.embeddedFORTH.de/temp/tts7.pdf
http://www.embeddedFORTH.de/temp/tts8.pdf
I am a bit sceptical about the usefullness of text-to-speech.
Steve Ciarcia ( who was involved with the Votrax 2 it seems )
published several articles in Byte with code in BASIC. That would
be a reasonable starting point for a simple implementation that
would probably work better then the CTS256.

MfG JRD
 
G

GdB

On 10 feb, 16:06, Rafael Deliano <[email protected]>
wrote:

Wow! That is A LOT of useful information, thanks a bunch! I know this
all can easilier and better be done these days, but it is the fun to
try and find all those very old IC's. And for some reason, especially
this CTS256... ;-)

[.....]
HM2007 4 channel filterbank speech recognition IC
        speaker dependent / trainable
Datasheet should be easy to find in www.
My board:http://www.embeddedFORTH.de/temp/hm2007p.pdf
My description ( in german ):http://www.embeddedFORTH.de/temp/hm2007p.pdf
The datasheet and the IC are slightly buggy.
It seems you added your board picture twice, instead of the
description... No big deal as I have never heard of this chip.

[.....]
Probably these:http://www.embeddedFORTH.de/temp/tts7.pdfhttp://www.embeddedFORTH.de/temp/tts8.pdf
I am a bit sceptical about the usefullness of text-to-speech.
Steve Ciarcia ( who was involved with the Votrax 2 it seems )
published several articles in Byte with code in BASIC. That would
be a reasonable starting point for a simple implementation that
would probably work better then the CTS256.
I know the TTS is very poor, but this is of course a chip from the 80-
ies, incredible by then! The wole idea is to try and find it.


Hertzlichen Dank! It's a wealth of information!

George
 
R

Rafael Deliano

I know this all can easilier and better be done these days,
Its a niche market semiconductor makers do no longer
address as there isn´t the market volume they nowadays dream
of.
There are neither modern small speech recognition ICs ( apart
perhaps from Sensory ). Nor are there speech synthesizers like
the digitalker that can serve as coprozessors to 8 bit
controllers and are supplied with prefabricated vocabulary
by the manufacturer.
In theory one can clone the oldtimers with modern micro-
controllers. But the required R&D high and there are no
obvious distribution channels.
SP0256 & CTS256 replacements seem to exist:
http://www.speechchips.com/shop/category.aspx?catid=3
But as said i am sceptical if text-to-speech really fits
typical embedded applications.
It seems you added your board picture twice, instead of the
description... No big deal as I have never heard of this chip.
Thats it:
http://www.embeddedFORTH.de/temp/hm2007.pdf
Available for $25 ( i got them via ebay ):
http://www.imagesco.com/speech/speech-recognition-technology.html
It's a wealth of information!
Make shure to store it, the files are only temporary on the website.

MfG JRD
 
G

GdB

Its a niche market semiconductor makers do no longer
address as there isn´t the market volume they nowadays dream
of.
There are neither modern small speech recognition ICs ( apart
perhaps from Sensory ). Nor are there speech synthesizers like
the digitalker that can serve as coprozessors to 8 bit
controllers and are supplied with prefabricated vocabulary
by the manufacturer.
In theory one can clone the oldtimers with modern micro-
controllers. But the required R&D high and there are no
obvious distribution channels.
Agree... I can not find it right now, but will post it if I can find
it, but I saw a project that uses the SP0256-AL2 ROM, but was
(MicroChip) PIC based. Funny...

SP0256 & CTS256 replacements seem to exist:http://www.speechchips.com/shop/category.aspx?catid=3
But as said i am sceptical if text-to-speech really fits
typical embedded applications.
I know that website, indeed they sell the SP0256-AL2 and, the much
more hard to find, SP0256-17. However, I'm actually looking for this
CTS256, where they sell a modern replacement (TTS256).

[....]
These guys are funny, $ 5 for sp0256-al2 DATASHEET!???
Make shure to store it, the files are only temporary on the website.
I did, many thanks.



Regards,

George
 
K

KR

The CTS256-AL2 is a text to code processor. It converts English text
to phoneme codes used by the
SP0256-AL2, which is a speech generator. It is all stuff from the 80-
ies, but it would be great fun to do a little project with this
chip... Any speech chips appreciated. Please let me know if you can
help me.

Thanks.

George

The other speech chip I recall was the SC-01. Had many pre-programmed
"speech" sounds - sound "parts" that could be played by the CPU in
strings to make pretty convincing "words" in english (and maybe some
parts of other languages if you wanted to try ?)

I saw this used was on several of the Gottlieb system 80(A) pinball
machines. (www.ipdb.org)

In the game "Caveman" (1982) it even synthesised very realistic
"cricket chirp / jungle" background sound effects as well as I vaguely
recall. (though they might not be that exciting by modern "sound
blaster" standards.)

Sound board in that was pretty basic 6532 processor system.
 
Top