Can u work out these audio eprom samples ?

[slider2732]

Here's a challenge for a Sunday !

I have some eprom sound samples from a drum machine and can't get them to open properly in Audacity/Wavelab/Goldwave/Wavosaur or anything else.
My wish is to burn some new ones for the unit, but they're not just 8-bit PCM at 18kHZ or something else easy enough it seems.

Drum unit is an MXR Drum Computer from 1983. They 'should' be approximately the settings as above, as that would correspond with similar units of the time 
Is it a compression algorithm ? are they encoded somehow ? Can you crack em ? !

Hope someone has gone down this road before and can help out

Have attached a zip with 3 eprom sounds, from 2716 eproms and a short mp3 of how wrong burned samples can sound..the obvious crappy one is supposed to be a TR-909 bass drum.

[RayB]

I'll give it a shot, but in my experience, many (most?) of the classic drum machines have had their samples dumped. Did you search the web for them?

[RayB]

Actually, it looks like you're on your way to finding a solution:

http://www.idmforums.com/showthread.php?t=16152

 

PS: Looking at the raw data of your KICK sample, I would guess that eprom had gotten erased, because you barely have any data in that one. (open it in notepad and compare it to the other two)

[slider2732]

Thanks Ray
I can vouch for the eproms still running, after dumping I put them all back in yesterday and had a mess around.
The kick would seem different...though hang on, i'll attach the original file in a zip. I just checked the original and it seems to be similar to the other 2. Not sure how that happened.

Indeed, posted over at the music forum too. Well spotted lol
Keith Barr (the designer of the MXR unit) has been very kind in emailing some details. He's thinking he altered data lines to make the circuit more elegant and used a ZIF to correct the alterations at his burner.
Checking the DAC (as suggested) didn't bring good news. All lines seem to correspond correctly. Following the DAC output, things go a little vague, with resistors  covering tracks etc up to final output.

If we could just emulate one of these files, to discover if they are indeed A-law or u-law, as Keith thought they may be, then we'd be on the better road.
All trials with various software at even creating files that look similar have proved fruitless.
Putting burns of those in the machine always ends up with a squark type noise, like a toy car horn.
In that, there may be a clue ?

Any mixed up data lines on the circuit board would still need data going in that was unaltered though, the altering coming after the eprom.

The processor is a Z80 and there is control code on a 2732.......Keith seems to find the project pretty cool, so I *think* it may be ok to at least email someone like yerself the control code he wrote. If, indeed, you know assembler (I dont). 

[slider2732]

Oh and yeah, I found the MXR samples, but they are 44.1kHz nice and shiny versions straight out audio from the machine. Same with TR-909, LinnDrum, Oberheim DX etc.
The .bin's don't seem to be around.

[ark_ader]

While we are at it, can we have some Samples for Starfire?

[slider2732]

Could be a fun aside, game samples on a drum machine hehe
Perhaps Starfire samples in a rising crescendo would be like it's sequel, Rolling Star Fire.

Normally there are encryption techniques for troublesome sets, includig arcade. But in the MXR's case, it seems it's down to a quirky method the designer used to read the samples.