> Those MP3s were created from 16 bit data, but once that data has been > transformed to the frequency domain as part of the encoding process, that > bit depth ceases to matter. In mpd's case, the MP3s are decoded as 28 bits > by libmad, and truncated to 24 bits. This results in theoretically lower > distortion than truncating or dithering that output to 16 bits. In your > case, by forcing 16 bit output with MP3 you are getting more distortion.
That reminds me: when the signal needs to be resampled because the output's sample rate doesn't match the file's sample rate (typical case for me: 44.1KHz CD input and an audio card that only supports 48KHz), would it be possible to do the resampling directly in the frequency domain as part of the MP3/Ogg decoding? It seems like it would be cheaper (CPU wise) and result in higher quality than using a separate resampling phase after decoding. Stefan ------------------------------------------------------------------------------ October Webinars: Code for Performance Free Intel webinars can help you accelerate application performance. Explore tips for MPI, OpenMP, advanced profiling, and more. Get the most from the latest Intel processors and coprocessors. See abstracts and register > http://pubads.g.doubleclick.net/gampad/clk?id=60134791&iu=/4140/ostg.clktrk _______________________________________________ Musicpd-dev-team mailing list Musicpd-dev-team@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/musicpd-dev-team