[pulseaudio-discuss] [PATCH 0/4] Add support for libsoxr resampler

[Alexander E. Patrakov]

12.11.2014 02:39, Andrey Semashev пишет:
> Hi,
>
> These patches add support for libsoxr resampler library to PulseAudio. The resampler is exposed in four selectable quality presets: lq, mq, hq and vhq (in the order of increasing quality). The patches update the code, build system and docs.
>
> libsoxr[1] is a resampler library extracted from the SoX utility[2]. It offers very good output quality and is very fast. Quality comparisons are available here:
>
> - http://src.infinitewave.ca/ - "SoX 14.4 High Quality" corresponds to hq quality. "SoX 14.4 VHQ Linear Phase" is close to vhq, the passband for vhq is slightly lower.
> - http://sox.sourceforge.net/SoX/Resampling - there are graphs for impulse response comparison with libsamplerate available at the bottom of the page.
> - http://lastique.github.io/src_test/ - I ran some more tests to compare with Speex resampler at different levels. Some performance measurments are also present at the end.
>
> In short, libsoxr is almost always faster than speex, and introduces much less distortions. Its passband frequency is slightly lower than speex though, and it can add a delay up to 20 ms in some cases.

We generally don't need a zoo of resamplers. But you have definitely 
changed something important from an earlier submission by Peter Meerwald 
so that the CPU figure became much better. I guess, that's 
SOXR_LINEAR_PHASE - that's the only obvious change.

Here are the facts, applicable to 44100 -> 48000 Hz resampling, as 
required by some sound cards when playing back CD rips:

  * Speex-float-5 never introduces audible distortions, even on 
specifically-crafted testcases at insane volume in an otherwise 
absolutely quiet room. So "even better quality" never makes sense - 
unless we are talking about non-human listeners.
  * Speex-float-1 (the current default) does not introduce audible 
distortions on music that you can buy in shops (or download), but does 
create audible artifacts on specially prepared test cases.
  * SOXR, even at its LQ setting (which, according to prior tests, is 
good enough) and with THIS set of patches, is indeed slightly faster 
than speex-float-1.

Statements about distortion audibility are based on the model described 
in this scientific paper:

http://www.mp3-tech.org/programmer/docs/6_Heusdens.pdf

Distortions caused by removal of high frequencies, as well as any other 
distortions, would have been counted as audible if a person could detect 
that in an ABX test. The nuance here (with "auditory masking" as its 
scientific name) is that the ear becomes less sensitive to very high 
frequencies when there is something else in the air - and in typical 
music, this "something else" definitely exists in a sufficient quantity.

Statement about speed is based on a direct CPU usage measurement by "top 
-d 10" on this CPU (forced to run at 933 MHz):

Intel(R) Core(TM) i3 CPU       M 370  @ 2.40GHz

Baseline (48 kHz playback without resampling): 0.7% (0.1% on null sink)
Speex-float-5 (44100 -> 48000 Hz): 5.6% (4.5% on null sink)
Speex-float-3 (44100 -> 48000 Hz): 4.3% (3.1% on null sink)
Speex-float-1 (44100 -> 48000 Hz): 2.5% (1.9% on null sink)
Soxr-hq (44100 -> 48000 Hz): 2.4-3.3% (1.6% on null sink)
Soxr-mq (44100 -> 48000 Hz): 2.5-3.1% (1.5% on null sink)
Soxr-lq (44100 -> 48000 Hz): 2.1-3.4% (1.7% on null sink)

One obvious thing here is that, while Speex-based resamplers yield 
stable performance on a real hardware sound card, Soxr CPU usage 
oscillates rather significantly. This effect does not exist on a null 
sink. Also, higher quality does not mean higher CPU usage, which is strange.

I will recheck the quality separately later today, in order to verify 
that it is still as good as in the previous tests. Please don't merge 
the patches until this is done.

-- 
Alexander E. Patrakov