You are not logged in.
Recently, albums I've tagged with Picard on OS X using the 0.11 UB have been causing iTunes to inexplicably distort my MP3s when imported into my Library. The distortions range from CD ripping-style 'blips' at the very beginning of a track, to the beginnings of tracks being cut-off, to the very ends of tracks skipping and repeating once.
I've tried to isolate exactly where/why the error is occurring in my tagging workflow. After Picard tagging, I've tested the problematic albums in a non-iTunes media player, such as Songbird or Quicktime, with no errors whatsoever. After drag & drop importing into iTunes, however, the tracks become distorted as described above, and will play distorted in other media players and my iPod, even after removing from iTunes. The distortions don't affect every tagged album, but I'd guess 50% or so.
So far I've tried:
- Deleting all Picard plugins and iTunes addons/scripts -- no change.
- Reverting back to previous versions of both Picard and iTunes -- no change.
- Complete uninstall and reinstall of Picard and iTunes, leaving both with complete default settings -- no change.
- Complete harddrive reformat and reinstall of OS X from scratch (no backup restore) -- no change.
I'm completely at a loss here. I've used the last two Picard Mac builds with no problem, and I'm not sure what could be causing this.
Any help would be greatly appreciated.
-Jesse
*Edit*
I created a waveform comparison using Audacity. The tracks are subtly but distinctly different before and after iTunes importing:
http://files.getdropbox.com/u/31934/wav … mation.gif
Last edited by jibegod (2009-07-23 03:21:01)
Offline
I think I see a time shift, as if iTunes import removed a small silence at the beginning of the track. Could you show us the same comparison zoomed on (for example) the 5 first seconds of the track? Please ensure that both waveforms start precisely at the beginning of the track and of course use the same level of zoom :-)
Offline
http://files.getdropbox.com/u/31934/waveform-zoom.jpg
Curiously, the program I'm using to view the waveforms (Audacity) isn't playing the iTunes-imported track with distortion, although it does play/show the post-import modification (time-shift).
Offline
I think silence removal should not remove from 0.20 to 0.30 but before 0.20. And also there is strange difference on 1.60. iTunes have options like cross fade, sound enhancer and sound check. Are you using any of these?
by the way are the wavs really 32 bit and not 16 bit?
Offline
I think you have several distinct issues: the time shift and the artefacts.
The time shift is obvious. iTunes "ate" roughly 15/100 seconds, but it did so by erasing what seems to be real sound. It seems that iTune simultaneously trims a part of the track and does a very fast fade in. I think the iTunes engineers thought this was a good idea. Indeed some mp3 readers (foobar2000 for example) do this when starting to read or when stopping. This avoids clicks. Doing it at the encoding step avoids doing it repeatedly each the track is read.
But this does not explain the blips. Do they sound like clicks (very very short, similar to what could often be heard on vinyls) or like artefacts (which may sound like bubbles or like very short whistle sounds)?
For those tracks where the end is repeated, do you hear the same thing in Audacity?
mp3 format is not always completely implemented by the mp3 readers, and encoders are known to do encoding mistakes. So either the iTunes encoder does something which is correct but uses a feature which is seldom implemented in readers, or there may be an encoding problem. The decoder included in Audacity may be smart enough to recognize the problem and correct it. What I find strange, though, is that iTunes seems to generate things which will he will trip on himself later.
Offline
@ym: The difference around 1.60 may be real but it could also be a false difference: sometimes removing a few milliseconds will change the aspect waveform at this scale because of "mistakes" in the routine which transforms the sample sequence into a graphical representation. The only way to be sure would be to zoom to the level where we see all the details of the waveform. As long as some parts of the waveform appear as envelopes only, it is difficult to be sure.
Last edited by davitof (2009-07-23 09:21:45)
Offline
I don't use any of the iTunes playback options (Crossfader, Sound Enhancer or Sound Check).
I tested the iTunes-imported track on my Windows 7 bootcamp install. The track played normally (no errors) in the latest version of Winamp, however the distortion was present on the iTunes Windows install.
I captured the 'blip' error from iTunes into a separate recording:
http://files.getdropbox.com/u/31934/track-capture.mp3
Offline
This is at the beginning of a track, right?
I am a bit surprised by what I see: I can clearly read LAME3.98.2 right in the middle of the sound frames. IIRC, you can find tags with text at the beginning and at the end of a mp3, not in the middle. I checked with my mp3 files and I see the same thing (I use the same version of Lame) for recent files, but for older ones this does not occur. It seems Lame 3.98 beta did not do this, but 3.98.2 does. I don't know enough about mp3 structure to say if this could cause the issue.
The blip is actually a scratch IMO :-D 5/100 seconds of completely saturated sound, this looks very much like a decoding+re-encoding issue. As we saw earlier, iTunes obviously does some editing to the files. I am guessing that instead of working directly on mp3 data (some editors are able to do this), it decodes the original, does whatever sound editing he wants to do, then re-encodes it. Maybe he changes the bitrate too? In this case, he couldn't do it directly in mp3 data, he'd have to decode-reencode. The problem seems to be during this step. Maybe he doesn't understand those strange bits of text inside the sound material.
Offline
Yes, the first 5 seconds or so. The file I linked is only captured audio from my sound card, not the original track itself, which I evidently saved with LAME 3.98.2. The original MP3 track which is being distorted by iTunes was actually encoded with LAME 3.97, with no errors. The 'cd scratch' sound you hear is entirely produced by iTunes, and does not exist in the original CD rip I made (FLAC).
So far, I've demonstrated that iTunes is directly altering sound data in my tracks. I'm not an expert on the data manipulation practices of audio player software, but this strikes me as a little disconcerting, with disturbing implications for the integrity of my entire music collection.
Last edited by jibegod (2009-07-23 11:56:40)
Offline
I don't have any opinion on this since I don't have iTunes. But now I feel even more happy about not using it :-)
To be honest, such a big bug seems strange. What kind of music do people listen to if nobody was bothered by this before? Or could it be that for some reason it happens only in your configuration?
Offline
I've been using Picard and iTunes together in OS X for 6 months and have tagged over 200 albums with no problems. My initial thought was that I had changed my system configuration somehow which introduced this playback error. However, in attempting to resolve my issue, I've formatted my harddrive and reinstalled OS X, and the errors still occur with a 'brand new' system config and freshly-installed Picard/iTunes. Also, as I mentioned before, I tested the iTunes-imported file (corrupted) under Windows, and iTunes read the 'damage' the same way my OS X iTunes install did, with the 'cd scratch' blip.
I suppose my next experiment will be tagging with Windows in a virtual machine..
Last edited by jibegod (2009-07-23 12:07:59)
Offline
Maybe something in the sound being compressed which generates a mp3 which itself gives the hiccups to the iTunes import decoder. I've had this strange thing: some of my tracks shut down my old mp3 player. As soon as I try to listen to one of those tracks, my old Archos just shuts down, immediately and brutally. I never investigated (they are tracks I don't listen to usually, so...) Do the errors always happen on the same tracks? If you reimport an old track, from the good 200 albums, do new errors appear?
Offline
First thing I'd try is setting Picard to write ID3v2.3 tags. I don't know if they fixed it, but iTunes used to not support ID3v2.4 properly and probably could misinterpret the data.
Offline
I've wiped my iTunes library database several times while trying to fix this, and have had to reimport my entire collection. As far as I can tell, none of my existing 'good' albums have manifested these errors where they did not exist before.
@luks: I've tried all combinations of ID3 settings in Picard; nothing appears to have an effect.
Last edited by jibegod (2009-07-23 21:46:31)
Offline
...and since ID3v2 tags are stored at the beginning of mp3 files, it could explain glitches at the start of a song. Are all artefacts at the beginning of the tracks?
Offline
They're all either at the very beginning or the very end (the repeated ending).
Offline
Ok, we identified a probable cause for the beginners. Now could you send a sample of the repeated end? Careful, since this is probably an encoding / decoding issue, I advise to use a tool such as mp3DirectCut
Offline
http://files.getdropbox.com/u/31934/tra … sample.mp3
This is a different song from a different artist/album than the earlier samples. This track also has the first second or two truncated from the beginning, but no 'cd scratch' blip.
http://files.getdropbox.com/u/31934/wav … sample.png
Comparison of the original track (top) and the iTunes-imported track (bottom).
Last edited by jibegod (2009-07-24 09:20:39)
Offline
Here is what I think is happening: iTunes shifted the whole track by 2 seconds But the new track was shorter than the old one, so iTunes added back the 2 last seconds of the old track. Plus one more second of silence. Why 2 seconds this time? Was there a 2 seconds silence at the start of the track?
Offline
I have a similar problem as jjbegod. I don't have any blips within the songs, but quite a few songs that have the last few seconds (about 3) repreated after the song has finished, before iTunes moves on to the next song. I have played one of the tracks with Preview too and the same problem occured.
My MP3s are tagged with ID3 v2.3. Let me know if there is any other information I can provide to help isolate the problem. I don't have any software to visualize wave forms though.
iTunes 8.2.1 (6)
MusicBrainz 0.11
OSX 10.5.8
Offline
Just an update: The issues I described appear to be resolved with Snow Leopard 10.6.1 and iTunes 9.0.1
Offline
