[pvrusb2] audio pops and cracks in the recordings

[Mike Isely]

On Tue, 23 Jan 2007, devsk wrote:

> > The pvrusb2 page doesn't mention the steps to get the v4l-dvb repository
> 
> ok, I figured that part and with the 'hg' I got the latest. The pops 
> are still there with the in-v4l pvrusb2. I couldn't modprobe the 
> latest standalone module because of unknown symbols although I did 
> compile it with V4LDIR set to point to the v4l-dvb sources.

Sorry, if I try to get into the business of documenting stuff that is 
not what I've created, then I'll spend my life writing documentation and 
not getting anything else done :-)  Knowing how to build v4l-dvb itself 
is best understood at linuxtv.org, though unfortunately the 
documentation there for that process is, well, uh, an opportunity for 
someone to contribute effort over there.  There are others in IRC in 
#v4l that can help with this sort of thing.  But I see you figured it 
out.  The only part I really described in the pvrusb2 docs was how to 
tie the standalone pvrusb2 driver into a v4l-dvb build (i.e. use V4LDIR) 
but even that isn't really needed since a roughly equivalent pvrusb2 
driver (minus the v4l-backwards compatibility pieces) is also in that 
repository.

If you are running a kernel older than 2.6.18, then you will get some 
undefined symbol warnings at driver link time, but that should not 
prevent you from being able to successfully use the driver.  (It's an 
issue caused by the kernel build system that is not really solvable 
outside of the kernel build system but it is also not fatal.)  If you 
are having modprobe problems with the standalone driver compiled against 
v4l-dvb, then well that should not be happening.  Best I can guess is 
that the v4l modules in your running kernel aren't the ones you built 
from v4l-dvb, or that you didn't completely build all the pvrusb2 pieces 
fresh when re-homing the pvrusb2 build on top of v4l-dvb, or that you 
need to rerun depmod -a to pick up newer classes of modules from 
v4l-dvb.  However, none of that is really important since the in-V4L 
pvrusb2 driver will work.

As of this moment, the in-V4L pvrusb2 driver still lacks the last round 
of changes (they haven't been picked up yet), but you can work around 
that if you need to:  Grab http://linuxtv.org/hg/~mcisely/pvrusb2 
instead of the main v4l-dvb repo (this is where I put my pending changes 
and where Mauro will usually pick them up from).


> 
> Another configuration difference could be that I have HZ set to 250. 
> How do these modules use HZ? I think most distros are using 1000. 
> Could this make a difference?

That is an interesting question.  I think I can provide a definitive 
answer:  If that were a real problem, I would expect it to impact actual 
playback of the audio stream (i.e. output through the sound card), not 
recording of it (i.e. creation and transport of the stream from the PVR 
USB2 hardware).  This is because the HZ value does not in any way have 
anything to do with that external hardware.  Also, if a too-slow HZ 
value were to cause the USB host controller to drop packets, then I'd 
expect severe audio AND video corruption, and I would expect it to be 
pervasive, not when there are "bursts" of audio (e.g. percussive noises) 
in the content.

Note also that by sending me that sample, you proved that the distortion 
is being introduced in the recording process not the playback side 
(since in that case the playback side was my laptop which otherwise 
renders audio just fine).  So I don't think the HZ value is a cause 
here.

I don't have any other ideas beyond me trying 2.6.19.2 or you trying 
2.6.18.1.  But I think that's a really low probability situation.  One 
inescapable fact is that so far you are still the only one our of 
presumably hundreds of people who are seeing this issue.  It's got to be 
something about your setup, but I'm not sure what.  If this really is 
some kind of driver or environmental thing, I'd sure like to figure it 
out because if that is the case then it's just a matter of time before 
someone else hits it.

  -Mike

-- 
                        |         Mike Isely          |     PGP fingerprint
     Spammers Die!!     |                             | 03 54 43 4D 75 E5 CC 92
                        |   isely @ pobox (dot) com   | 71 16 01 E2 B5 F5 C1 E8
                        |                             |