LMC-TRANS@GUNTER-ADAM.ARPA (04/12/86)
From: LMC-TRANS@GUNTER-ADAM.ARPA Has anyone considered putting speech in amigaterm. Page 125 of the May issue of Computer Shopper has a simple program that converts an ASCII text file to speech. Why couldn't that feature be put into Amiga term, so that we could listen to out E-mail. Seems interesting. Any comments?? Vaughn Wasem -- lmc-trans at Gunter-adam -------
mykes@3comvax.UUCP (Mike Schwartz) (04/14/86)
A long time ago, in a galaxy far far away, I posted the first of the CLI replacements (MyCLI). After posting it, I made some modifications to it to make it say out lout whatever came in over the wire. It was kinda neat being able to hear my Amiga read net.micro.amiga to me out loud. However, there are some major problems with using speech in this manner. 1) Spoken voice is very slow. Byt his I mean that even 1200 baud is . much faster than the voice can keep up with. AmigaTerm has a . few problems, most importantly the fact that it has too small . a buffer and will overrun if it gets too many bytes too fast. 2) When reading the news, you have to listen to the Amiga read all . the mail headers (how do you pronounce UUCP - the Amiga doesn't . say it the same...). 3) The translator/narrator are very good, considering their size, . but they are not good enough. For example, they will say each . digit in the number 100 (one-zero-zero) instead of (one hundered). By the way, I have looked at trying to fix the buffer problem with Amigaterm, but with no success. I have tried increasing the read buffer size (in the rs232 parameters), but there is no buffer big enough. I tried using x-on/x-off handshaking (I use AmigaTerm to talk to the vax here) and setting the IO Control character double word to appropriate values, but it doesn't work. As a matter of fact, at 9600 baud (connected to my IBM PC), using the CTTY command of MS-DOS, you can actually control the PC from the amiga. However, if you type a long file, AmigaTerm gets a buffer overrun after a while and starts printing garbage on the screen. 300 baud is the only speed I have found that seems to work (it failed at all the rest...). It seems to me that in order to get speech to work well on the Amiga, you should use a tool like speechtoy to determine what misspellings you need to make to make the narrator say what you really want. Then you chould take the translated phonem string and embed it in your 'c' source and call the narrator only. This makes the speech always correct, and does not require the unneeded overhead of the translator library.
michaelk@copper.UUCP (Michael Kersenbrock) (04/16/86)
Some years ago when I bought my Votrax type 'n talk text to speech synthesizer, I mentioned in net.something_or_other that I "listened" to usenet articles with some of the same facination (and problems) that have been mentioned already. I had also suggested another version of all newsgroups, the variation for this newsgroup would be net.micro.amiga.gab (netname).gab where folk would write articles with a more phonetic spelling. It seemed that this would make prow nun c a shunz by speech synthesizers happy and spelling errors unnoticeable. Although it could be argued that some net groups already spell "that" way, I got a sense of little enthusiasm O WAY L ! Any interest now? :-) -- Mike Kersenbrock Tektronix Software Development Products Aloha, Oregon
eblack@well.UUCP (Eric Black) (04/16/86)
In article <1837@caip.RUTGERS.EDU>, LMC-TRANS@GUNTER-ADAM.ARPA writes: > From: LMC-TRANS@GUNTER-ADAM.ARPA > > Has anyone considered putting speech in amigaterm. > Vaughn Wasem -- lmc-trans at Gunter-adam I have been using my version of AmigaTerm (I call it AmyTerm) which has the following features: 1) spawns a subtask to translate & narrate incoming text (optionally) 2) use audible or visible bell 3) gives 24 lines of 80 chars (yes, 80) 4) same X-modem as Amigaterm (i.e. checksum only), but slightly better aesthetically 5) many more menus 6) line settings including parity, echo, CR/LF handling goodies 7) can send a BREAK! 8) uses CONSOLE device to emulate ANSI terminal behavior 9) uses INTUITICKS interrupts and queries serial device to send incoming chars to console in batches (much faster than 1 at a time!) I'll be posting it soon, I expect. I'd like a few more people to play with it for a while... The speech aspect is indeed nice for calling in to BBS's (or even to work) and do other things while stuff comes in. There is definite room for improvement in its throughput for translation, however. Right now what I do is double-buffer input, and send X-OFF when the second buffer is full and the first hasn't been returned yet from the translator/ narrator subtask. Could be better... I'm using it now to post this message; "coming soon in a theater near you". ---- Eric Black "Garbage in, gospel out." UUCP: {sun,pyramid,hplabs,amdcad}!chronon!eric WELL: eblack BIX: eblack
local-info-amiga-request@ics.UCI.EDU (04/21/86)
From: Michael Kersenbrock <copper!michaelk@caip.rutgers.edu> Some years ago when I bought my Votrax type 'n talk text to speech synthesizer, I mentioned in net.something_or_other that I "listened" to usenet articles with some of the same facination (and problems) that have been mentioned already. I had also suggested another version of all newsgroups, the variation for this newsgroup would be net.micro.amiga.gab (netname).gab where folk would write articles with a more phonetic spelling. It seemed that this would make prow nun c a shunz by speech synthesizers happy and spelling errors unnoticeable. Although it could be argued that some net groups already spell "that" way, I got a sense of little enthusiasm O WAY L ! Any interest now? :-) -- Mike Kersenbrock Tektronix Software Development Products Aloha, Oregon