Glen,
Thanks for posting that link here (in this thread, for completeness). That's the link that had been broken for a long time which the fellow who runs VT100.net just fixed a couple days ago after I wrote to him. It's a great manual and explains all the escape codes (for controlling non-spoken features) and the phonemic instructions (ones in [square brackets] ) that are spoken. It also provides the functions of the dip switches on the board (for setting baud rate/data bits and some phone line functionality options.) It was also amusing to find a short list of literature references listing works written by the linguist who did the algorithm development.
John,
No specific plans yet, but I have always had some interest in speech synthesis since we got a TI-99 speech synthsizer in 1984 or so. I have played with the TI TMS5220 speech chip which speaks by being sent an LPC10 data stream. However, one needs a library of LPC10 phoneme data to assemble the data stream, and then on top of that, you need a text to phoneme algorithm. In the TI-99, the phoneme data was on a ROM in the speech synthesizer sidecar and the text to phoneme algorithm was in the Terminal Emulator II cartridge. When I played with the TMS5220 on its own, I used a PC to send the LPC10 stream, built from phoneme data for a limited number of full words that was also stored in (and had been copied out of) the TI-99 speech synthesizer rom. I did obtain and install, but never tried to make my own LPC10 streams, with a Win3.1 application for digitizing speech from the PC microphone, processing, and making an LPC10 file.
At any rate, all off that does not make for a compact, stand alone text to speech device. The DECtalk is perfect for that. I was thinking I could put this on the pdp 8/e, so that when playing Adventure, the DECtalk can read the instructions out loud. I may also play a bit with the voices other than Perfect Paul, or perhaps seeing if I could work out a custom voice that I liked.
Lou
Bookmarks