Re: [xwax-devel] Working on JACK code

[Mark H. <ma...@po...>]

On Tue, 4 Nov 2008, Tyson Henning wrote:

>> Here is mine: Compiled this last version of xwax with jack. Results:
>> Compiled and running ok. Only one thing: some the track accelerate or
>> downrate a little bite... I tried with two turntable.
>> Might be related to my fresh installed system, though. Gonna give deeper
>> look once I got the time to. (I still haven't back a really stable
>> rt-kernel... )
>
> Actually yes, now that Yves mentions it the wow is noticable. I've
> picked this up in a number of vinyl control implementations, both
> commercial and open source though... It seems to be an approximation
> problem - running average of N samples flickers by up to a percent
> based on a number of factors, most likely the major influence is the
> shocking amount of handruff on my Serato records. : ) But seriously I
> think that sort of thing is where the issue stems from, but it's not
> specifically an xwax problem, it's a problem with vinyl control in
> general - determining how fast a platter is spinning within what,
> maybe a thousandth or five-thousandth of a revolution is a non-trivial
> operation, bound to get some inaccuracy.

No, it sounds more lie this is likely to be dropping of input audio, 
resulting in skips in the timecode signal. So larger buffers are needed.

Unfortunately if you are using JACK, xwax does not know about this loss of 
audio. But if you are using ALSA, then you will see an 'xrun' message to 
the console.

If you aren't using a realtime kernel then you will be unlikely to achieve 
small buffer sizes required for xwax to work at its best, and make sure 
xwax is running with permission to run it's audio thread with realtime 
priorities.

> Perhaps smoothing would fix this, though of course that stuffs 
> scratching pretty badly.

There's already smoothing in the appropriate places, although this can be 
tuned. In the case above, the problem is more severe caused the loss of 
audio.

-- 
Mark