I've read:
Perhaps I'm reading out of context, but say I am playing an MP3 file(ReClock audio-only mode) -> wasapi-> headphone jack out-> headphones
Sampling rate: 44.1 KHz
Resolution: 16 bits
..should I be setting FFDShow Audio to output in 32-bit float and ReClock in 24-bit int padded? Or more like just both at 16-bit?
1) DAC's are 24bit at best...drivers will either ignore the 8 bottom bits, or worse convert from 32 to 24.
2) no point in doing useless conversions, Reclock will either do the same or better conversion to 32float than ffdshow would...just make sure to decode lossy audio to 32float whenever possible, in order to avoid rounding errors and useless conversions.
You should input in reclock the format that will need less prior conversions. For example:
(1) a 24 bit lossless soundtrack should be sent to reclock as PCM 24, If you convert it to 32 bit floating point and input that in reclock, it would not be lossless anymore, because ffdshow's conversion to 32 bit floating point is not as accurate as reclock's conversion, even though you would not be able to hear the difference.
(2) a lossy encoded soundtrack decoded using 32 bit floating point, should be sent to reclock as floating point. ffdshow's conversions from floating point to PCM are not as accurate as reclock's
*ALWAYS* decode lossy audio to 32float whenever possible(read above for the reason )
Perhaps I'm reading out of context, but say I am playing an MP3 file(ReClock audio-only mode) -> wasapi-> headphone jack out-> headphones
Sampling rate: 44.1 KHz
Resolution: 16 bits
..should I be setting FFDShow Audio to output in 32-bit float and ReClock in 24-bit int padded? Or more like just both at 16-bit?
Last edited: