Skip to main content

Topic: Audiotester picky: what is wrong about these files? (Read 3143 times) previous topic - next topic

0 Members and 1 Guest are viewing this topic.
  • Porcus
  • [*][*][*][*][*]
Audiotester picky: what is wrong about these files?
The following files (< 600 kilobytes total) are fine with foobar2000, but not with VUplayer's "Audiotester.exe" tool. And curiosity gets the better off me.
https://www.sendspace.com/file/ao5g63
https://www.sendspace.com/file/oxql65

(Sendspace files can be downloaded many times, right?)

They were generated from the file found at http://freemusicarchive.org/music/Aidan_Baker/Live_in_Moscow/ . 100 megabytes, so I cut it down to the few first seconds using mp3directcut with the same result: Bad ID3tag, and when I remove the tag (using mp3tag), then audiotester reports a loss of sync at the beginning of the file.


  • j7n
  • [*][*][*][*][*]
Re: Audiotester picky: what is wrong about these files?
Reply #1
Mp3GuessEnc finds "unexpected data at 90160", length 1044 bytes, in the original Aidan_Baker_-_01_-_PMT60_playmytape__Dom_Club.mp3 between the ID3v2 tag and the audio datŠ°. This corresponds to 1 audio frame at 320 kbit/s. It seems that the encoder reserved this space for writing a Xing frame at the end of the encoding process, but it was never written. Most players are robust enough to silently skip over unknown data in MP3 files. You can cut this empty space out with a hex editor or 'repair' the MP3 with a stream rebuilder/header tool.

  • Porcus
  • [*][*][*][*][*]
Re: Audiotester picky: what is wrong about these files?
Reply #2
Thanks for the  tip about that utility.

It seems that the encoder reserved this space for writing a Xing frame at the end of the encoding process, but it was never written. Most players are robust enough to silently skip over unknown data in MP3 files. You can cut this empty space out with a hex editor or 'repair' the MP3 with a stream rebuilder/header tool.

Yes, fb2k's stream rebuilder fixes it. But not its header repair functionality. fb2k uses ffmpeg for a bit of things, and ffmpeg demuxing does not help either; so those tools seem to consider it part of the stream and not part of the headers, right?

Not that I bother so much about it - fb2k never saw anything wrong and I never heard anything wrong - but one does get curious.

  • j7n
  • [*][*][*][*][*]
Re: Audiotester picky: what is wrong about these files?
Reply #3
I've seen mp3 files without the xing frame (and without gapless data) occasionally before. It looks like Mp3DirectCut copies everything before the first frame as the ID3 tag without parsing it. Demuxing an elementary stream doesn't make sense; it is already fully demuxed. I would expect video tools to retain junk data when demuxing from a container because this data represents time and sync with video. Fixing of the resulting stream is up to other tools.

  • Porcus
  • [*][*][*][*][*]
Re: Audiotester picky: what is wrong about these files?
Reply #4
"Issues" (for curiosity; non-issues for listening purposes) at the end of files.  fb2k gives no errors nor warnings.

Both are from the same source, Dutch radio offering festival podcasts, I guess they convert everything using the same tool.  Tagless.

One of them: last frame is allegedly "corrupted".
Code: [Select]
First frame found at 0 (0x00000000).

Frame 147359 at offset 92385071 appears to be incomplete.
  294 bytes available, 627 expected.
  This frame is corrupted.

Detected MPEG stream version 1 layer III, details follow.
  File size                : 92385365 bytes
  Audio stream size        : 92385071 bytes
  Length                   : 1:04:09.378 (3849.378 seconds)
  Data rate                : 192.0 kbps
  Number of frames         : 147359
  Blocks per frame         : 4 (granules per frame 2, channels per granule 2)
  Audio samples per frame  : 1152
  Audio frequency          : 44100 Hz
  Encoding mode            : stereo
  Min global gain          : l= 99  r=100
  Max global gain          : l=210  r=210
Flags
  Error protection : no
  Copyrighted      : no
  Original         : no
  Emphasis         : none

Block usage
  Long block granules   :  566160 (96.1%)
  Switch block granules :   14858 ( 2.5%)
  Short block granules  :    8418 ( 1.4%)
  ---------------------
  sum                   :  589436

Ancillary data
  Total amount   : 299018 bytes (0.3%)
  Bitrate        : 0.6 kbps
  Min packet     : 1 bytes
  Max packet     : 592 bytes
Max reservoir                          : 334 bytes
Scalefactor scaling used               : yes
Scalefactor selection information used : yes
Padding used                           : yes

Frame histogram
 192 kbps : 147359 (100.0%), size distr: [  9022 x 626 B, 138337 x 627 B]

 0 header errors.

Maybe this file is encoded by Helix

The other, here it is the second-to-last frame, "incomplete" but "valid", whatever that means:
Code: [Select]
First frame found at 0 (0x00000000).

Frame 134790 at offset 84505078 appears to be incomplete.
  417 bytes available, 627 expected.
  This frame is valid.

Detected MPEG stream version 1 layer III, details follow.
  File size                : 84505495 bytes
  Audio stream size        : 84505495 bytes
  Length                   : 0:58:41.071 (3521.071 seconds)
  Data rate                : 192.0 kbps
  Number of frames         : 134791
  Blocks per frame         : 4 (granules per frame 2, channels per granule 2)
  Audio samples per frame  : 1152
  Audio frequency          : 44100 Hz
  Encoding mode            : stereo
  Min global gain          : l= 76  r= 70
  Max global gain          : l=210  r=210
Flags
  Error protection : no
  Copyrighted      : no
  Original         : no
  Emphasis         : none

Block usage
  Long block granules   :  517626 (96.0%)
  Switch block granules :   13898 ( 2.6%)
  Short block granules  :    7640 ( 1.4%)
  ---------------------
  sum                   :  539164

Ancillary data
  Total amount   : 2284 bytes (0.0%)
  Bitrate        : 0.0 kbps
  Min packet     : 1 bytes
  Max packet     : 18 bytes
Max reservoir                          : 334 bytes
Scalefactor scaling used               : yes
Scalefactor selection information used : yes
Padding used                           : yes

Frame histogram
 192 kbps : 134791 (100.0%), size distr: [  8252 x 626 B, 126539 x 627 B]

 0 header errors.

Maybe this file is encoded by Helix

  • Last Edit: 05 May, 2017, 04:07:15 AM by Porcus