Audacious

Why don't you enable the resampling effect as the message suggests?

John Lindgren wrote:

Why don't you enable the resampling effect as the message suggests?

Sorry for the late reply. The thing is that resampling spends more cpu resources. Without it the cpu usage of audacious is tiny. On an older, slow machine, high quality resampling with crossfading doesn't even work as expected. The lowest resampling quality setting doesn't sound too good. I'm not sure about the quality settings in between. But it would still be nice if resampling could be disabled and just have crossfading when possible.

I'm surprised that you would notice the extra CPU load. Most Linux systems nowadays are configured to do some kind of software resampling anyway, so whatever extra CPU is taken by Audacious should be made up for by savings at the system level (provided Audacious is resampling to whatever rate the system is using). On my machine, enabling the linear interpolation method within Audacious actually uses slightly less overall CPU time than leaving ALSA to do the resampling.

The trouble with just disabling the message is that you will still lose audio data when the song changes. If you had the crossfade overlap set to 10 seconds, then the last 10 seconds of the song (which were buffered in order to perform the crossfade) will be cut off when the next song starts.

Ok, so I guess losing audio is not worth it. Thanks for everything.

I don't notice much difference in cpu usage between the resampling qualities src-sinc-best-quality and speex-float-10. Pulseaudio seems to use speex-float-3 by default, these days. But what causes increased CPU usage for me is setting the rate to 48000. I don't suppose it makes much difference to human ears between 48000 and 44100, but that's just a guess.