Vintage sound chips make cool music, but what's hiding inside of them? Which algorithms are they using?
In the past years I've been dedicating my free time to building emulators for old digital sound chips from synthesizers/game consoles, in an effort to preserve them and make them usable without the original hardware. During this time I went through lots of interesting stuff: silicon reverse engineering, logic analyzers, weird compression schemes, forgotten audio algorithms and more. In this talk I want to share with you what I learned from this process and what's next. Specifically, I will explain how I'm currently reverse engineering a custom DSP from the 90s just by using an Arduino Mega and lots of speculation, and how you can probably apply the same process to other chips as well.
This talk is kind of a sequel for my previous 38C3 talk "Proprietary ICs and dubious marketing claims - let's fight these with a microscope!", where I reverse engineered an old Roland digital piano from silicon die shots, since it was using some custom algorithms no one knew about.
Instead of analyzing fixed behavior hardware, this time I want to focus more on DSPs: processors that can execute code to manipulate audio data in real-time. Since some of them are not documented, the bytecode for their programs will look like random binary without a datasheet.
Instead of going through the silicon directly, I was able to figure out a lot about them just by probing them, building a pretty accurate emulator without destroying the original chip.
Licensed to the public under https://creativecommons.org/licenses/by/4.0/