TI-99/4A Speech Demo II

Another speech demo. The speech chip used in the TI-99/4A speech synthesizer was designed around 1976-1977. A trimmed down version was used in TI's famous Speak and Spell toy.

Here, I took a section of a pod-cast from www.retrobits.com (great site!) and downsampled to 8khz mono, then processed in QBOX.

You can see the assembly source code if you change the view to full screen ;-)

The original WAV file was 283KB, the speech data, after processing, was 4.6KB :-)

NOTE: What you are hearing here is an EMULATOR, not the real thing. It sounds better on the real thing :-)

Also, it's much more readble if you view this video in High Quality.

Hope you retro fans like this vid ;-)

Category:

Entertainment

Tags:

• TI-99/4A
• ti99
• ti-99
• speech
• retro
• synthesis
• synthesizer

[yeowza9]

Woah. I never realized the TI-99 had such a clear voice synthesizer. I think it can still complete with most of today's voice synthesizers. It's clear and you can understand it.

[MarkWillsUK]

Hi. Yes I am continually amazed at it. And also at the level of compression. The data that does the talking is 98% smaller than the WAV file that I used as a source! Glad you liked it!

[orion1052003]

Aw, heck, son! That was cool! Email me. Will there be a speech tutorial for Win99/4A users? How do I do what you did, and go beyond the limits of the resident list of words in the speech synthesizer. So far I haven't been able to understand CALL SPGET. Also can't use Terminal emulator and XB at the same time.

[MarkWillsUK]

He he! Speech tutorial? Your wish is my command! Just click on the video response above and enjoy!

Regards,

Mark.

[duftmand]

scary...

[metabog]

Nice work dude, wish I could take a better look at the code.

[MrChiknman]

Wow! It almost sounds kinda real! :D

[gjc82071]

@Serpico261 That's what I said........

[douro20]

@orion1052003

The speech is synthesized using phonemes and prosodic information contained in the code. It is not being generated using a dictionary.

[douro20]

@gjc82071

This is synthesized speech created using phonemes and prosodic information generated from an actual recording.

[douro20]

The chip is an 8-bit DSP which uses LPC-compressed diphone samples. The diphones can either be stored on the chip's internal ROM or loaded from memory. The phonemes, and accompanying prosodic information, are streamed to the chip, which it uses to generate the speech by concantenation of the compressed diphones.

[Serpico261]

hmm, that was a synthesized speech? i have problems to believe that..its sounds so human.

[gjc82071]

@yeowza9 Something sounds kind of fishy with that demo. That doesn't sound like a "speech" program, (like Windows Narrator) as much as it does just....."sampled" voice, some how played back, via the TI99, utilizing a low sample rate. It sounds "weird", cool, amazing, but weird. Like a "real" person.

[neons99]

I don't understand... why did you start with WAV file if this is true speech synthesis????