Re: [ecasound] hq-resampling with ecasound ( now what ? )

Hi Kai,

thanks for looking into the issue. it is not a necessary thing for me right now, so
there is no need to push
it, though it would be nice to be operative so that I could continue my research &
development of Hi-End
linux based music systems.

Kamil

Kai Vehmanen wrote:
> On Mon, 20 Jun 2005, Kamil Wencel wrote:
>
>> #> jackd -R -dalsa --device hw:1 --rate 96000 --playback --nperiods 8
>>
>> #> ecasound -B:rtlowlatency -b:128 -f:32,2,96000 -i
>> resample-hq,44100,'audiofile.wav' -f:32,2,96000 -o jack
>> #> ecasound -B:rtlowlatency -b:128 -f:32,2,96000 -i
>> resample-hq,44100,'audiofile.wav' -f:32,2,96000 -o alsa,rme96
>
> [...]
>
>>
>> But, as soon as I plug jack inbetween it took me hours to get any
>> sound out of it. And when it finally did it sounded
>> worse than the old broken ecasound upsampling. Instead of fine music
>> you get your music again overlayed
>> by noise and scratches.
>
>
> Yes, there seems to be a problem. The resampler produces different size
> blocks if the blocksize is not an integer multiple of the resampling
> ratio (128/(96000/44100)), and jack output object can't handle this (not
> a problem for alsa and file outputs). This problem becomes obvious when
> you play with the jackd blocksize - try first with "-p 128" and then "-p
> 2048", and you will certainly hear the difference (note with jackd, you
> need change jackd blocksize, ecasound's -b:xxx is always overridden by
> the jackd system-wide blocksize).
>
> OTOH, when resampling between the "easy ratios" (44100<->22050,
> 48000<->96000, etc), the output will be fine. But unfortunately
> 96000/44100 is not one of these cases. And just be sure, this problem
> only affects JACK input/outputs.
>
> I'll add a runtime warning about this problem to the CVS-tree. It's bad
> if you get audible artifacts without any warnings.
>
> To solve this problem, one would need to add an intermediate buffer to
> the resampler, and thus add latency to the processing chain. Not
> impossible, but trivial either --> to the todo-list.
>
> --
> links, my public keys, etc at http://eca.cx/kv
>
>
> -------------------------------------------------------
> SF.Net email is Sponsored by the Better Software Conference & EXPO
> September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices
> Agile & Plan-Driven Development * Managing Projects & Teams * Testing & QA
> Security * Process Improvement & Measurement * http://www.sqe.com/bsce5sf
> _______________________________________________
> Ecasound-list mailing list
> Ecasound-list@email-addr-hidden
> https://lists.sourceforge.net/lists/listinfo/ecasound-list

-------------------------------------------------------
SF.Net email is Sponsored by the Better Software Conference & EXPO
September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices
Agile & Plan-Driven Development * Managing Projects & Teams * Testing & QA
Security * Process Improvement & Measurement * http://www.sqe.com/bsce5sf
_______________________________________________
Ecasound-list mailing list
Ecasound-list@email-addr-hidden
https://lists.sourceforge.net/lists/listinfo/ecasound-list
Received on Tue Aug 23 20:15:06 2005