The documentation doesn’t seem to specify if the multi-byte samples (16, 24, 32) are big or little-endian.
Does this depend on the underlying raw data format (like aiff will be big endian and wav will be little) or does FMOD guarantee one or the other when using the Sound::readData api?
When using the Sound::readData multi-byte audio will be returned in the native endian of the platform regardless of the file format. Since there are no big-endian platforms (at the moment) Sound::readData always returns little-endian. For formats like AIFF we will perform the endian swapping if necessary to ensure little-endian is returned.