Re: Potential P2 in Audacity 2.0.0 alpha
Gale Andrews <gale <at> audacityteam.org>
2012-02-03 17:01:40 GMT
| From Steve the Fiddle <stevethefiddle <at> gmail.com>
| Fri, 3 Feb 2012 13:09:51 +0000
| Subject: Potential P2 in Audacity 2.0.0 alpha
> This relates to bug 152 http://bugzilla.audacityteam.org/show_bug.cgi?id=152
> and this forum topic: http://forum.audacityteam.org/viewtopic.php?f=16&t=63406
>
> Summary:
>
> Very high sample values in audio data can cause Exported audio to be
> corrupted from the position of very high sample value for the
> remaining duration of the file.
> The corruption in Audacity is persistent until Audacity is restarted.
>
> Steps to reproduce:
>
> Dither set to "shaped".
> 32 bit float format audio track containing 1 or more samples with
> values of 100 dB or more (NAN values may also cause this problem, but
> I've not fully tested that).
> Export track in any integer format.
>
> Test the Imported track and it will be found that it flat-lines at -1
> dB from the position of where the very high sample value was.
>
> Create a new Audacity project without restarting Audacity.
> Record or generate some audio.
> Export in any integer format.
> The Exported file is corrupted (flat-line at -1 dB).
>
>
> Gale wrote here:
> http://forum.audacityteam.org/viewtopic.php?f=16&t=63406&start=20#p169975
> "1.3.3 processes the project correctly, but I haven't tried later
> versions. I'm not surprised, because I've seen several other reports
> of "flatlining" similar to this on the Forum recently and I didn't
> think they were all inability to open the source data (though one
> was)."
>
> So it appears that this is not an isolated incident, and it does not
> matter if the "invalid" samples are caused by an improper Nyquist
> command (as in bug 152) or by a corrupt sample value occurring during
> recording (as in the forum topic).
>
> For testing purposes, simple way to create an "illegal" sample is to
> run the following code in the Nyquist Prompt on one or more samples of
> an audio track:
> (snd-const (db-to-linear 100) 0 *sound-srate* (get-duration 1))
>
> Is there any way that audio track samples can be limited to sensible
> values, for example, less than +50 dB?
> Can this be fixed before 2.0?
>
> I think that this bug must be at least P3, but really it seems more
> like a P2. Can anything be done before 2.0?
>
> (this e-mail was written in a bit of a hurry as I think you all need
> to be aware of this asap, so apologies for any errors)
Of course I'd like it fixed but I was envisaging entering it as P3 then
raising it and some others to P2 after 2.0 so as to prioritise them
(or some other way to do so).
The problem has been around since at least 1.3.8 (just tested).
I can only recall one other confirmed Forum report like this before
(though I cannot find it right now). In that case the recording had a
very obvious visual and audible glitch where there were very high
sample values. Exports were then corrupted at -1. But in the current
case the fault is not obvious, so cases may be more common than we
thought.
I'm not sure why there have increasing numbers of reports of "flatlined"
exports recently. It could merely be that I've been actively looking for
them, thinking they could be project reopening bugs. These sort of
reports look to be fairly common on the Forum, but given the reports
are usually vague they are usually dismissed as user running out of
disk space and not noticing the holes in the project, or having a
fragmented drive.
Gale
------------------------------------------------------------------------------
Try before you buy = See our experts in action!
The most comprehensive online learning library for Microsoft developers
is just $99.99! Visual Studio, SharePoint, SQL - plus HTML5, CSS3, MVC3,
Metro Style Apps, more. Free future releases when you subscribe now!
http://p.sf.net/sfu/learndevnow-dev2
RSS Feed