Sound differences in GSF -> ROM conversions? by SoundGLSI at 2:14 PM EDT on May 14, 2026
This is something I'm interested in knowing if anyone noticed it. I was listening to the Castlevania Circle of the Moon rip, and something that caught my attention is how different the "Requiem" theme sounds in emulators and GSF playback. At first I thought that was something related to emulation, but that wasn't the case. I grabbed gsfopt to convert the minigsf track to a ROM file. And guess what? Requiem sounded like it was being played at a lower sample rate. I don't know how the game does it, but in emulators, it sounds clearer. Even in ancient versions of mGBA. I re-ripped the game with saptapper to see if it was the rip, but no dice. same result.
Are you referring to the difference between the original, unmodified game ROM and the GSF? If so, I was able to confirm that as well. That's interesting.
In the past, there have been a few games, such as Custom Robo GX, where rips created with Saptapper were incorrect. It might take a little time, but I'm interested in looking into this in more detail.
Edit: Just like in Custom Robo GX, I've confirmed that m4aSoundMode is called specifically for streamed tracks in this game as well.
Yes, I'm refering to the differences between the GSF rip and the unmodified ROM. Thanks for confirming it. So, there are games that change the sample rate in ssome songs? That certainly would explain the behavior with the current GSF rip. Harmony of Disonance has some streamed tracks with higher sample rate too.
Circle of the Moon specially increases the playback rate by passing 0xC0000 to the m4aSoundMode function just before streaming begins. The updated GSF has already been uploaded and should be released at a later date.
The same applies to Harmony of Dissonance. While the reverse engineering is complete, I haven't had time to update the GSF yet, so it may be a little while before it's available.
For accuracy's sake, let me clarify: while previous GSF implementations had lower mixing rates, the new version uses a much higher mixing rate based on analytical results. By specifying a higher rate, the resulting sound will likely contain fewer high-frequency components—this isn't a quality degradation, but rather a reduction in noise.