russell@ihuxu.UUCP (Larry Russell) (12/02/84)
I would not think that the SID chip has sufficient filtering capability to synthesize recognizable speech. There is, however, a speech synthesis board available (at Toys 'R' Us among other places) which I have not actually heard myself -- Larry Russell AT&T-Bell Laboratories Naperville, IL ihnp4!ihuxu!russell
wjr@utcs.UUCP (William Rucklidge) (12/04/84)
> I would not think that the SID chip has sufficient filtering capability > to synthesize recognizable speech. On the contrary, the SAM speech synthesizer program produces easily recognizable speech using only the SID chip. It does turn the VIC chip off while speaking, to eliminate processor freezes, but this can be disabled and the voice quality deteriorates only marginally. The program requires no extra hardware and takes up a small amount of BASIC memory, and adds a few new commands to BABASIC, notably "SAY". Text to be spoken can be written either as phonetics,`or as normal English spelling (with the "Reciter" loaded), and SAM does its best (quite good) to pronounce English's irrational spelling system. The voice's tone, speed and quality can be changed easily - there is even a demonstration program that sings "The Star Spangled Banner" (atrociously, but the tune and words are fairly clear), and recites several well-known speeches. This message brought to you with the aid of the Poslfit Committee. "You can always put something in a box." William Rucklidge University of Toronto Computing Services {cbosgd,decvax,harpo,ihnp4,utcsrgv,{allegra,linus}!utzoo}!utcs!wjr -- This message brought to you with the aid of the Poslfit Committee. "You can always put something in a box." William Rucklidge University of Toronto Computing Services {cbosgd,decvax,harpo,ihnp4,utcsrgv,{allegra,linus}!utzoo}!utcs!wjr
ma175xby@sdcc3.UUCP ({|stu) (12/05/84)
> I would not think that the SID chip has sufficient filtering capability > to synthesize recognizable speech. > There is, however, a speech synthesis board available (at Toys 'R' Us > among other places) which I have not actually heard myself > -- > > Larry Russell > AT&T-Bell Laboratories > Naperville, IL > ihnp4!ihuxu!russell I've heard quite understandable speech on the c-64 without any additional hardware though I do not remember the name of the package. I have the Magic Voice module put out by commodore and it sounds quite nice. It features a female voice. Some drawbacks are that it has a limited vocabulary of about 200 words and the mystery of programming more words has yet to be discovered. Matt Buse UCSD
spencer@garfield.UUCP (Spencer Barnes) (12/06/84)
> I would not think that the SID chip has sufficient filtering capability > to synthesize recognizable speech. > There is, however, a speech synthesis board available (at Toys 'R' Us > among other places) which I have not actually heard myself > -- > > Larry Russell > AT&T-Bell Laboratories > Naperville, IL > ihnp4!ihuxu!russell I can think of one trivial example off hand. A newly released game (at least in this area) called GHOSTBUSTERS has a reasonably recognizable synthesis of one word..... THATS RIGHT, YOU GUESSED IT, IT'S GHOSTBUSTERS!!! While I haven't fully figured out how they did it yet, the basis seems to be precise adjustments to the white noise voice. Anybody with a better handle on the method pls let me know, 'cause, if "they" can do it, WE CAN TOO!! Spencer (WHO YA GONNA CALL) Barnes.
mab@druxp.UUCP (BlandMA) (12/06/84)
> A newly released game (at least in > this area) called GHOSTBUSTERS has a reasonably recognizable synthesis of one > word..... THATS RIGHT, YOU GUESSED IT, IT'S GHOSTBUSTERS!!! > > While I haven't fully figured out how they did it yet, the basis seems to be > precise adjustments to the white noise voice. Anybody with a better handle > on the method pls let me know, 'cause, if "they" can do it, WE CAN TOO!! The noise waveform should be able to generate most all of the non-vocal phonemes (s, t, k, ch, ...). 'k' is easy - use low attack and decay values, maybe 0 for attack and 2 or 3 for decay. An 's' probably would have an attack of 2 or 3 (I'm guessing at these values, since I'm not near my 64 right now). If you're not into POKES, it's fairly easy to experiment with the SID values using the music programs that are on the market. I was able to make a human-sounding 'b' sound the other night using the triangle waveform and a filter. Seems like it was something like attack 3, decay 4 or 5, low sustain, low pass filter on, and the cutoff frequency was somewhere around 1500-2000. The frequency of the note was in one of the middle octaves. As the frequency changes, the cutoff frequency also has to change. I'm still new at figuring out how the filters work, but in this case, the filter is what makes the attack sound like a human-like 'b' rather than just a synthesized music note. Now I'm getting curious. When the Ghostbusters bass line music was posted last week, I went home and put it on my 64. Now it looks like I'm going to have to add speech to it. If I'm successful, I'll post something, but don't hold your breath waiting. -- Alan Bland {ihnp4, allegra}!druxp!mab AT&T Information Systems Labs, Denver
dave@zinfandel.UUCP (12/15/84)
Actually, SID *can* produce very good speech. I base my judgement on a
product called S.A.M (Software Automatic Mouth). It uses allophones, or
you can just type in regular English and SAM will make some assumptions.
You can also control inflection. It's amazing how inflection can improve
the quality of speech -- no more 'monotone'. Questions even sound like
questions.
I was impressed with it, especially since it is 100% software. I think
it sells for about $45 or so, depending on where you look. Most of the
mail order houses carry it.
Dave Funk
Zehntel, Inc.
2625 Shadelands Drive
Walnut Creek, CA 92598
(415) 932-6900 x309
....!decvax!sytek!zehntel!dave
...."zehntel!dave"@BERKELEYengle@loral.UUCP (Jack Engle) (12/18/84)
> Actually, SID *can* produce very good speech. I base my judgement on a > product called S.A.M (Software Automatic Mouth). It uses allophones, or > you can just type in regular English and SAM will make some assumptions. > You can also control inflection. It's amazing how inflection can improve > the quality of speech -- no more 'monotone'. Questions even sound like > questions. > > I was impressed with it, especially since it is 100% software. I think > it sells for about $45 or so, depending on where you look. Most of the > mail order houses carry it. > > > > Dave Funk > Zehntel, Inc. > 2625 Shadelands Drive > Walnut Creek, CA 92598 > (415) 932-6900 x309 > > ....!decvax!sytek!zehntel!dave > ...."zehntel!dave"@BERKELEY *** REPLACE THIS LINE WITH YOUR MESSAGE ***