[Flac] Rather serious flac problem

[Free Lunch]

Hi Kerry,

On 9/16/05, Kerry Hoath <kerry at gotss.net> wrote:
> There is no standard for wav64 and a .wav in its standard form can only hold
> 2gb of data.

In this case the WAV files were less than 2 GB.

> libsndfile can handle larger .wav files and so can other audio applications
> but usually nonstandard subchunk data means that the wav reader in flac
> can't understand the file.

This was simply a recording which was terminated before completion so
the final size of the WAV was not written in the header.

That is not an uncommon event during content creation.  When reading a
WAV, many programs check the file size against the header and will ask
whether to use or ignore the header info.  While that sort of
interaction may not be desireable wihtin flac, I think the default
behavior could be more robust (even if it just exits with an error).

> Many audio programs put comments and all sorts of other metadata in .wav
> files.

Yes but not in this case.

> I agree flac should probably copy unknown subchunks uncompressed to the
> output stream but i'm not sure how doable this would be given the current
> flac decoders and the like.
> it is also probably a good idea to test the exit status of *all* envokations
> of flac to catch warnings since I assume a warning returns a non-zero error
> code but I am not totally sure of this.

Unfortunately, the return code in this case is 0.  So as a bug, it is
very serious.


Regards,

  Brian


> ----- Original Message -----
> From: "Free Lunch" <freelunch at gmail.com>
> To: <flac at xiph.org>
> Sent: Friday, September 16, 2005 10:17 PM
> Subject: [Flac] Rather serious flac problem
> 
> 
> Okay.. I love flac but just had a rather serious failure that really
> shakes my confidence.  It resulted in the near loss of a master audio
> recording.  Fortunately, I have a backup.  Though there may have been
> other cases where I have lost original material because I have been
> compressing a lot of originals and deleting them after doing a 'flac
> -t' on them.
> 
> Basically, flac failed with a 'warning' but still created flac
> archives which tested ok.
> 
> Here's the scoop:
> 
> % flac *wav
> 
> flac 1.1.1, Copyright (C) 2000,2001,2002,2003,2004 Josh Coalson
> flac comes with ABSOLUTELY NO WARRANTY.  This is free software, and you are
> welcome to redistribute it under certain conditions.  Type `flac' for
> details.
> 
> options: -P 4096 -b 4608 -m -l 8 -q 0 -r 3,3
> umb.1.wav: WARNING: skipping unknown sub-chunk 'p9'
> umb.2.wav: WARNING: skipping unknown sub-chunk 'c=¶'
> % flac -t *flac
> 
> flac 1.1.1, Copyright (C) 2000,2001,2002,2003,2004 Josh Coalson
> flac comes with ABSOLUTELY NO WARRANTY.  This is free software, and you are
> welcome to redistribute it under certain conditions.  Type `flac' for
> details.
> 
> umb.1.flac: ok
> umb.2.flac: ok
> % ls -lt
> total 3105736
> -r--r--r--  1 root root       4186 Jun  5 00:43 umb.2.flac
> -r--r--r--  1 b    b    1933467692 Jun  5 00:43 umb.2.wav
> -r--r--r--  1 root root       4186 Jun  4 22:37 umb.1.flac
> -r--r--r--  1 b    b    1243668524 Jun  4 22:37 umb.1.wav
> 
> % file *
> umb.1.flac: FLAC audio bitstream data, 24 bit, stereo, 44.1 kHz, length
> unknown
> umb.1.wav:  RIFF (little-endian) data, WAVE audio, Microsoft PCM, 24
> bit, stereo 44100 Hz
> umb.2.flac: FLAC audio bitstream data, 24 bit, stereo, 44.1 kHz, length
> unknown
> umb.2.wav:  RIFF (little-endian) data, WAVE audio, Microsoft PCM, 24
> bit, stereo 44100 Hz
> 
> 
> % flac -v
> flac 1.1.1
> % uname -a
> Linux foobar 2.6.7-gentoo-r11 #12 Fri Aug 12 19:40:17 EDT 2005 x86_64
> AMD Athlon(tm) 64 Processor 3000+ AuthenticAMD GNU/Linux
> 
> 
> My guess is that in this case, the wav file header may not correctly
> reflect the length of the wav file.  I would think that flac could be
> coded to handle these cases correctly.
> 
> 
> 
> Thank you!
> 
>   Brian
> _______________________________________________
> Flac mailing list
> Flac at xiph.org
> http://lists.xiph.org/mailman/listinfo/flac
> 
> 
>