Whats the optimal or acceptable format/quality of the music files?
I had some success using PCM signed 8bit stereo
its not a deal-breaker using the current format, unsure about the workload it may require for you to allow for all "Mono" and "Stereo" 8bit or 16bit 44Khz 22Khz and 11Khz. Could the audio format be detectable or set through an argument?
I personally would be a fan of the lower quality formats (currently using 200MB in 8bit Stereo format after converting all original Carmageddon audio tracks).