First post, by Jonas-fr
Owner of a SC-55 here, I want to shrink down my setup (486 computer hooked to a SC-55 via MIDI) while retaining the SC-55 abilities. I'm currently focusing on getting a 1:1 copy of the SC-55 soundfont (I won't go in the debate between what's the closest soundfont today ,there's some very good clone/reproduction more or less faithful today but that's not the point of this topic). I'm wondering whether it would be possible to extract the exact sounds contained in one (or two) of the chips of a SC-55 to get a soundfont and then flash it on a Dreamblaster X2.
Of course if such dump was easy I think we already got one today they shouldn't be so many reproductions so I'm wondering whether the following will be possible (see it as an thought exercise) :
- Dumping the actual digital bytes of the SC-55 soundfont from a real unit (in contrary of dumping the analog sound signal from each instrument as it is done today to make soundfont reproductions)
- Extracting then converting the soundfont from Roland Sound Canvas VA to a format that could be flashed on a Dreamblaster (SF2 or DXB)
Regarding the first option we can try to do that via software (a glitch) to extract needed data, or by hardware by snooping signals with a logic analyser (non destructive method) or by boiling the needed chip(s) in acid a dumping by microscope the actual sound tables (tedious but it has been done on other hardware).
The first method (software) would needs an exploit which need some fairly good understanding of the underlaying firwmare of the SC-55. The second method (logic-analyser) seems to be the least hard one provided that we find a data line between soundbank memory and the CPU to be poked. The last method (acid then microscope analysis) would prove to be the hardest of the bunch and needing the more resources (time, talent, hardware).
Regarding the second option it would implies that Roland actually got the real soundfont for a SC-55 (and not a reproduction) actually flashed/burned in the factory while the SC-55 were made. If that's true then we "only" need to extract it from the software and then convert it to an usable format (SF2).
Well this is it, I certainly missed details here or oversimplified things by being naive but please bear with me and do not hesitate to chime in with your advices and/or view on the subject. And if feel up to join a reverse engineering effort on this matter (hardware of software) please chime in too 😀