[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: Teleport use of Votrax speech synthesizer - how to get it to work?
On Tuesday, November 13, 2012 3:25:16 PM UTC-6, BLuRry wrote:
> On Tuesday, November 13, 2012 12:31:54 PM UTC-6, BLuRry wrote:
>
> > On Thursday, November 8, 2012 11:47:33 PM UTC-6, Hot Rod wrote:
>
> >
>
> > > Hi,
>
> >
>
> > >
>
> >
>
> > >
>
> >
>
> > >
>
> >
>
> > > Long story short - I ended up looking a bit at the game Teleport, by Mike Abbott and published by Cavalier Computer, 1982. Recognizing that Mike Abbott's other games had secret keys in them, and not seeing Teleport on my list, I thought I'd poke around in it.
>
> >
>
> > >
>
> >
>
> > >
>
> >
>
> > >
>
> >
>
> > > I did find that it does have a few. But what I also learned is that apparently this game is meant to use something called a Votrax speech synthesizer (best I can tell is that this was some sort of phoneme-based speech chip). The interesting thing is that the phoneme text and the code to play the text is present, but I don't know what to get it to play through. AppleWin and Mockingboard emulation doesn't seem to be enough. What would this have played through in 1982?
>
> >
>
> > >
>
> >
>
> > >
>
> >
>
> > >
>
> >
>
> > > Some of the secret keys are for playing secret messages; I can read the phoneme text for them in the code, but it would be neat to be able to play them (some look humorous). I can't think of any other secret keys like these, that lead to audio messages. There are also regular game messages to be played too.
>
> >
>
> > >
>
> >
>
> > >
>
> >
>
> > >
>
> >
>
> > > Seems like a Mockingboard would be the nearest thing. Any ideas on how to map it in?
>
> >
>
> > >
>
> >
>
> > >
>
> >
>
> > >
>
> >
>
> > > ]HR
>
> >
>
> >
>
> >
>
> > Well, nevermind... found a hi-res decap photo here:
>
> >
>
> >
>
> >
>
> > http://www.visual6502.org/images/pages/Silicon_Systems_SSI_263P_die_shots.html
>
> >
>
> >
>
> >
>
> > So it's possible to extract the rom from the image. looks like 29 bits wide by 66 (or so) rows -- that would be the phonenme characteristics.
>
> >
>
> >
>
> >
>
> > -B
>
>
>
> I was off a little... 64 rows (one per phoneme I'd imagine) and 27 bits across. I enhanced the rom and color coded the blobs to see the data better. Here's the rom decoded, though it might be flipped 180 degrees the wrong way (so the last row might be phoneme #0, etc). If the blobs are 1 and the absence of a blob is zero then this is the SSI-263 rom decoded:
>
>
>
> 101000011000101000001000011
>
> 111100111001110001011000000
>
> 111100111000111010010001001
>
> 111100111000110011000010011
>
> 111100011001010011001001000
>
> 111100111001110001011000001
>
> 111100111000110001010011011
>
> 111100111000101000010000001
>
> 111100111000001001011001010
>
> 111100111000001001001000010
>
> 111100111100000001000001001
>
> 111100111100000001001001010
>
> 111100011101001011000001011
>
> 111100010101011011011001011
>
> 111100010101001011011001011
>
> 111100010101011011001011001
>
> 111100010101001011010000001
>
> 111100110000001001001001011
>
> 111100110000001001000011011
>
> 111100110100000001010000000
>
> 111100110000010011011000001
>
> 111100110000011001000010000
>
> 111100110000000010001000011
>
> 111100100000100011000001001
>
> 111100110100010011000001000
>
> 111100110101000001011001010
>
> 111100010101001011010001010
>
> 111100010100011011000011010
>
> 111100100000011001001001000
>
> 111100100000000000001001011
>
> 111100100000100000011000010
>
> 111100010001011010011001001
>
> 111100010001100011011010011
>
> 111100110000110001000001011
>
> 111100110000101001000010011
>
> 111100110000000000001001001
>
> 011100110000100000010000011
>
> 011100111000100001000000011
>
> 111100111000000010011000001
>
> 011011010010001100010100000
>
> 010011011010101101000100010
>
> 011010011010000000011000001
>
> 101100011001101010001000011
>
> 001100111001101010001010011
>
> 101011011000101000001000011
>
> 001011011000101000001000011
>
> 101100011001101000101000111
>
> 110001010000100100011011001
>
> 110011010010110100010100010
>
> 111001011000100101011010010
>
> 111010011010100101011000010
>
> 111000010010000010011011010
>
> 111010010010000000011001010
>
> 110000010000110101011010011
>
> 110010011000001101000000001
>
> 111000110000000010110001110
>
> 111000111000100000100001100
>
> 111000011001110001101000100
>
> 111000111000001001001000011
>
> 111000011001000010001001000
>
> 111000110000010010010001011
>
> 111000111000000011000000010
>
> 111000010001011011011010010
>
> 111000110000100001000000011
>
>
>
> Given the rest of the data sheet and a more careful study of where the rom sits in the chip it should be possible to match up each bit column with the registers/filters that the bits control to produce somewhat of a better simulated ssi-263 sound. It will not be possible to produce a 100% accurate reproduction since the chip synth is analog, but hey this is a start...
For anyone trying to use the data, it's upside-down (so rotate the whole thing 180 degrees or just invert it). I don't think there's any necessary LSB or MSB in the data as much as most bits are used to control on/off of different filters. I'm trying to gauge if any other parts of the chip produce a fixed waveform, there are some slight irregularities that look almost like waveforms that are stored in-chip as segments of varied lengths. Since the photo only shows one layer it's hard to tell though.
-B