Convolution filtering

[simoncn]

The convolution file must have the same sample rate as the file being transcoded. If you provide a convolution file with a different sample rate, FFmpeg resamples it internally before using it for convolution. I would expect slightly better results from avoiding this resampling but I haven't done any tests to compare.

The special options to reduce the gap (mentioned in your other post) compensate for the convolved output being slightly time delayed compared with the input. Assuming there is a delay of 32768 samples, the input is padded with 32768 samples of silence at the end and the output has 32768 samples removed from the beginning. The remainder of the file is processed unchanged, so audio quality is not affected. I don't know whether 32768 is always the correct value to use.