TI-99/4A Speech Demo
Вставка
- Опубліковано 11 жов 2007
- Demo of the TI-99/4A talking. It's always been difficult to generate speech on the TI using the TI Speech Synth, because the synth chip uses LPC (linear predictive coding) to produce the tones and sounds. This demo was produced QBOX Pro (a very old piece of 16 bit Win 3.0 software). I recorded my voice as a WAV, down sampled, removed breathing noises, corrected BIAS, and processed using QBOX. QBOX gives you the data to fire at the synth chip. I then wrote a simple machine code program to do the talking and update the display. Enjoy. The TI-99/4A emulator you see here is Win994A by Cory Burr. MESS produces (arguably) slightly better speech results. This program DOES work on a real TI too!
QBOX is (formerly) commercially available software written by the designers of the TI Synth chip themselves (after they left TI). The software was re-discovered by Ben Yates after his endless search of the internet trying to find suitable software. Without Ben, it would still be impossible to generate ones own speech with the TI speech synth. There are no instructions with QBOX, but Ben kindly supplied copies of his correspondance with the programmers, which contained some rudimentary instructions. Thanks Ben. :-)
If anyone is interested in the souce, I have published it (TMS9900 assembler) here:
www.planet-99.net/wotw.htm - Розваги
That was really interesting. I felt compelled to find and download this QBOX program and start messing around with it myself.
Great idea to use the LPC synthesis with pre-processed audio. Cheers!
Dear Mark. BRAVO! Fantastic work! I love how you´ve explored a work around for us who wants to find an Original 1980 (Mother LPC chip) from Texas Instruments, the TMS5100 chip, but fail. Let me just correct you: This is not a speech synthesizer speaking. It's known sequence = the LPC code (Coding) - it´s fed a sentence and determines thru a 10 band digitally quantized "vocoder" how to make it sound with heavy distortion but intelligibly OK. Synthesizer simulates a vocal tract's formants. Cool!
derocov I still can't believe my dad Ben Yates made this!
this is historical if you think about it..never before was there anything like this...i like the voice on here better than the voice windows gives you! CLASSIC!
Hello! I'm very proud of what my dad did, and thank you for uploading this it reminded my dad its name because somehow he forgot
Back when I was in Junior High we had a class where a TI with the speech synth was set up in the back of the room, of course every chance we had when the teacher wasn't looking we had it saying ALL kinds of things they never intended!
We had a TI at the back of the class in Junior High, when the teacher wasn't paying attention we had it saying ALL kinds of things I'm sure the designers never intended it to!
We had one of these in a class back when I was in school, and believe me, once we figured out how to make it talk we had it saying all kinds of things Texas Instruments never intended!
naughty. i like it
Nice. Good choice of literature. Sounds like War of the TI Worlds, cuz...how did you get the speech so good? I have only used the limited word list. How does assembly access the speech sytnhthesizer?
Thanks for saving this software from obliteration! :)
where you can find QBOXPRO software? thaks for your valuable work . regards
whoami999ster whoami999ster thanks to Ben Yates actually, he was the one who made it if you check the description :) actually he's also my dad! He always forgets about it tho lol
I have managed to get the processed speech data to be spoken with the Arduino Talkie library. However I had to delete the first character from the data (0x60). What is the format?
I think in the game Parsec they did it like this with the woman's voice. This would be so interesting for me! The sound of the voice here is fascinating, and even more must the few data be, that comes out when you convert wav to lpc. i'd really like to get more special information about this
Thanks for posting the code. What does the circuit look like?
Hmm it was pretty good.
Whoa! It sounds just like Steven Hawkins!
That sounds a bit like Speak & Spell, only I think Speak & Spell sounds better.
i have a ti-99/4a how can i connect it to windows based pc to do what you did, or HOW DID YOU do that?
Hoping for a tutorial to be written on this :)
@MarkWillsUK
You mean you didn't generate the speech using diphones?
@lachlant1984
It uses the same chip family being used in the Speak'n'Spell.
You have a new email :D
No-one told me we were about to be attacked by Martians - how long have we got left, it takes weeks to get here doesn't it? This wasn't on the news this morning, I must get a gun or a spear or something, do you know a good shop? Christ I'm scared, anyone have a shelter? are they like big radioactive dudes or green and slimy like that guy next door? help someone for god's sake help. Or we could just mate for the last time now - on the lawn outside my house...help someone I need help...
This is not votrax. It's pure audio recorded from a human being in bad khz.
dude you dont need a really old tv you can just get an adaptor from radio shack for like 3 bucks and use any tv.. oh yeah!
200 pepole have a TI994A, 1080569594 pepole have a win994a
Weird, the ti-99a does a better job of text to speech than windows 7 does.
Sounds better than Microsoft Sam! How come I can acually hear expression and even an accent in this old program? Most text to speech even today sounds pretty monotonous but this doesnt, how does it know exacltly what sort of way it should speak it?
Because it's not text-to-speech. The author recorded his own voice, fed it into a program which converted the WAV for use by the TI speech synthesizer chip, and wrote a program which played back that data.
The TI-99/4A _could_ do text-to-speech, with some software cartridges like _Terminal Emulator II_, but it was nothing like Microsoft Sam -- it was very "robot-like".
Email