You might want to check on AtariAge. There is a thread there with the source code to Parsec in it--and that has code for direct control of the speech synthesizer in it. . .I don't remember which thread it was off-hand, but it was from sometime last year (with the last updates early this year).
07-28-2014, 10:15 PM (This post was last modified: 07-29-2014, 10:18 PM by sjt99.)
(07-19-2014, 09:14 PM)ksarul Wrote: You might want to check on AtariAge. There is a thread there with the source code to Parsec in it--and that has code for direct control of the speech synthesizer in it. . .I don't remember which thread it was off-hand, but it was from sometime last year (with the last updates early this year).
Attached is a demo speech cart I threw together with data from Atari's SW speech ,from the arcade game. Since the TI 994a and SW used the same speech chip, thought it should work and it does.
08-30-2014, 12:25 AM (This post was last modified: 08-30-2014, 01:14 AM by sjt99.)
Someone, reminded me of an old xb program that used unique speech.
I set out to see if I could incorporate unique speech data to used by XB as they did years ago.
I wrote a conversion program to create the tokenized xb raw data strings from data I had previously produced using qbox pro.
These data lines are viewable in the XB program as a character representations of the data between 0 and 255 sent to the speech synthesizer. These lines also include the necessary header and ending strings for the synthesizer to work. I kept speech data to 240 bytes /2 chunks. This size was easy to work with and is well within the max string size allowed for xb speech and split for the data lines.
I've attached a sample program.
It will run under win994a and most of time produces the correct speech, not always though.
I've included 2 different methods that will produce the same result in the program.
I like this. The Speech Synthesizer is one of those great peripherals that didn't get its due back in the day. It is good to see that development for it continues.
(08-30-2014, 07:24 PM)ksarul Wrote: I like this. The Speech Synthesizer is one of those great peripherals that didn't get its due back in the day. It is good to see that development for it continues.
I feel the same way.
I remember that Bert/Ernie demo and though intrigued by it I had no idea how the raw data was made.
I didn't even know that originally the program included the data in a separate file.
The author, never took credit.
At the time I think a lot of people thought that someone had taken the resident vocabulary words and altered them to sound like Bert & Ernie.
Highly unlikely.
Most likely it originated with TI itself, using their speech development system.
I have given thought about updating the Bert & Ernie program with the rest of the dialog that was left out. Now that I have worked out settings for Bert & Ernie that sound acceptable.
Actually, you can generate really good LPC speech using either the Text to Speech disk, the Speech Editor cartridge, or the Terminal Emulator II cartridge. It even describes how to save the data files and modify them for deeper or higher range voices--and then import them into other applications, including BASIC.
(09-03-2014, 11:27 PM)ksarul Wrote: Actually, you can generate really good LPC speech using either the Text to Speech disk, the Speech Editor cartridge, or the Terminal Emulator II cartridge. It even describes how to save the data files and modify them for deeper or higher range voices--and then import them into other applications, including BASIC.
I think I have the editor manual, I'll have to get a fresh look at it.
Thanks for pointing that out.
This speech data was created using original recorded speech and optimizing it as LPC data to produce the closest representation of the actual original recorded speech source. Hopefully I'll get better at it.
Ti created their speech for cartridges using their hardware and software system, from what I understand. (marketed for like $10000) They also had a linguist involved to help produce the best quality speech possible.
BTW I tweaked the 5220 table in qbox to more closely match the 5200 chip used in the Speech Synthesizer.
I believe Stuart Conner has one of their speech development systems, now that I've thought a bit on this subject. Data keeps bubbling to the surface from conversations I've had with various TI people over the last 30 years. . .
(09-05-2014, 08:11 PM)ksarul Wrote: I believe Stuart Conner has one of their speech development systems, now that I've thought a bit on this subject. Data keeps bubbling to the surface from conversations I've had with various TI people over the last 30 years. . .
Hi, Guest
