[sdiy] Decoding Kawai K3 Waveform ROM's

[Bob Weigel]

It'll be interesting to see what you do with this.  The K3 was my first 
synth.  Sold it years ago for a Korg Wavestation actually.  Never been 
sorry about that :).  The kawai was a good machine to learn some things 
on.  THe brick wall at around 10K frequency kind of limits it's sound 
quality though.   And of course I needed a much more flexible machine to 
begin sequencing my projects etc.  The WS is a marvelous sound design 
machine for it's day.  The lack of interesting filters being it's main 
drawback really...well along with difficulty to expand waveforms.  But 
the rest of it is really great.

Anyway the higher harmonics are totally useless because of the cutoff.  
I got partway into building an expansion for memory on mine then sold it 
for the WS.   I glanced at the patent....without spending all afternoon 
trying to figure it out...sounds like this was a very simple idea of 
dealing with processing limitations of the day by having a second 
harmonic processing element dealing with only odd harmonics which are 
not 2x multiples of the existing harmonics in the other processing 
system.  Anyway, it well may be something of this is related to what you 
are seeing there since the patent did just preceed the release of the 
K3...sometime in 1986 and maybe too early to involve this at all.  Not 
sure.  I got mine in '87 when they were trying desperately to get rid of 
the thing that had been around too long.  Hehe.  They still didn't give 
me a huge break on it but the old lady next door's son ran the store and 
I decided to chunker down on it. -Bob

Plutoniq9 wrote:

> *This didn't seem to post to sdiy when i posted to ah last nite.
>
> Hi,
>
> I've been interested in creating a new waveform ROM set for the kawai 
> K3/m for quite a while now and have been making great progress the 
> last few days.  However, it uses a method of storing waveforms in a 
> very odd fashion;
>
> - Uses 2 ROM's (#09 & #10) for waveforms, but uses them in 
> combination/parallel (not just to extend memory space).
>
> - ROM #09 is 8-bit Unsigned INT 32,768-bytes. Rom#10 is 8-bit Signed 
> INT 32,768-bytes.
>
> - ROM #09 is 32768-bytes x 8-bit. Each of the 31 waveforms occupies 
> 1024-bytes of space. Within this 1024-byte space, 5 different octaves 
> of the waveform are "interleaved", higher octaves contain less sample 
> data (i.e 512-bytes, 256-bytes, 128-bytes, 64-bytes, 32-bytes x 2). 
> This is to avoid aliasing I imagine.
>
> This much I have figured out. I wrote a small c++ program that has 
> allowed me to extract the "base" 512-byte wavefoms from ROM#09. Next 
> step here will be to go the other way, convert a 512-byte single-cycle 
> waveform into the K3's 1024-byte interleaved format.
>
> HOWEVER........
>
> The second ROM (#10) is still a mystery to me....... The only hunch I 
> got is that is may relate to this patent;
>
> http://free.patentfetcher.com/GetPatentPDF.php?f=Pats/US/45/26/US4526081.pdf 
>
>
> But I'm not sure I understand it well enough. If anyone can offer some 
> help, it would be greatly appreciated! Ultimate goal is to create 
> replacment Kawai K3 wave ROM's containing "cream of the crop" 
> single-cycle waveforms (i.e wavestation, prophet VS and various 
> "classic" analogue synth waveforms from real machines etc.)
>
> To help, i've created a zip file containing the sine waveforms (both 
> de-interleaved and interleaved) from ROM # 9, and the "mystery" 
> 1024-byte sine from ROM # 10. I'll also include the .bin dumps of both 
> waveform ROM's. The big question is......
>
> What the fudge does ROM # 10 have to do with ROM # 09? Is it to extend 
> harmonics like in the patent # 4526081?
>
> Zip file can be found at audiosyn.com/files/K3%20mystery%20waveform.zip
>
> Thanks
>
> Plutoniq9
>
> _________________________________________________________________
> The average US Credit Score is 675. The cost to see yours: $0 by 
> Experian. 
> http://www.freecreditreport.com/pm/default.aspx?sc=660600&bcd=EMAILFOOTERAVERAGE 
>
>
>