[Ffmpeg-devel] alsa input / output

Måns Rullgård mans
Wed Mar 21 01:52:47 CET 2007


Michael Niedermayer <michaelni at gmx.at> writes:

> Hi
>
> On Wed, Mar 21, 2007 at 01:15:51AM +0300, Vladimir Mosgalin wrote:
>> Hi Rich Felker!
>> 
>>  On 2007.03.20 at 16:18:19 -0500, Rich Felker wrote next:
>> 
>> Well.. sorry for OT but I really think that your point on this is VERY
>> wrong.
>> 
>> > Of couse, as an aside I strongly question the usefulness of lots of
>> > this "functionality". It seems more like bloat to me. If the hardware
>> > doesn't support a particular samplerate it should simply require the
>> > application to provide a supported samplerate, rather than doing
>> > horribly-painful-to-listen-to nearest-neighbor resampling in the
>> > library level...
>> 
>> That's exactly the problem you get with device interface; it is just
>> supposed to "work". But with alsa accessed through library interface,
>> you can do anything you desire, when your applications doesn't support
>> this. And you mean application should _always_ use built-in
>> resamplers? THAT is the right way to get painful-to-listen-to sound,
>> because most of resamplers will suck. With alsa, however, you just
>> choose whatever you want, depending on your CPU and personal choice
>> (believe it or not, but SRC_SINC_FASTEST sounds better than
>> SRC_SINC_BEST_QUALITY and all other SRC_SINC libsamplerate filters to
>> me).
>
> so the low quality resampler from libsamplerate sounds better to you then
> the high quality ones from libsamplerate, while i dont have much faith
> in libsamplerate i cant belive that their fast low quality code sounds
> better then their hq code, maybe your audio hardware is broken and the
> additional low pass filtering done by low quality resamplers mask that
> your furher comment below points strongly in that direction, only very
> odd hardware is limited to 24bit/sample
>
> besides this it would be interresting to run proper double blind tests
> between libsamplerate and the lavc resampler, if libsamplerate beats us
> it would be trivial to use their filter, after all samplerate conversation
> is just a trivial linear equation with coefficients choosen to please
> the latest audophile trend

Assuming the input is not undersampled, is there not exactly one
mathematically correct resampling?  It's been a while since I studied
signal processing, so I may have forgotten something.

BTW, ALSA 1.0.14-rc3 can use libavcodec for resampling.

-- 
M?ns Rullg?rd
mans at mansr.com




More information about the ffmpeg-devel mailing list