As part of a larger project involving procedurally generated music, I modified blargg’s GameMusicEmu test program for easier integration with other audio programs. You can find the program on GitHub as vgm2wav.

There already exist a great many audio conversion utilities for video game music files to WAV, MP3, or MIDI, with varying degrees of sophistication. The vgmtrans project does conversion in a sophisticated way that (I am guessing) selects the MIDI instruments on the output from the set of voices specific to each game developer. I couldn’t get it to work properly and it would crash a lot. This was probably my fault since I didn’t exhaustively debug my Frankenstein build environment.

Most other programs are GUIs that can run under Wine in Mac OS X. For my purposes, a GUI is too slow.
Many examples can be found on Zophar’s Domain.

I was after a simple program where I could specify an input file and optionally send the output to stdout for use with programs like WaoN or Sound eXchange (SoX) in shell or Python scripts. I needed this functionality to process large numbers of songs in a fairly fast and standardized way.

By building on the great library and example programs that blargg put together, I was able to get a nice tool working fairly quickly. It also gave me a chance to brush up on my C skills–the last time I did anything significant in C was about ten years ago! Doing this also made me appreciate the hard work and cleverness that has gone into emulator development.

Testing has been limited but a full SPC/NSF to MIDI with WaoN seems to work fine for the NES and simple voices on the SNES (especially with voice isolation).

Check out the project on GitHub and let me know if you found it useful, or if you have more questions!