CD TOC Disc ID and/or Free DB computation bug: Last audio track before data track is 2:32 too long

Continuing the discussion from Fate of the old forums:


Old FreeDB imports (and maybe Picard and isrcsubmit DiscID as well? I don’t remember) used to have a bug that made the last audio track of an Enhanced CD show a length 2 minutes and 30 seconds longer than its real length.

Example:

Edit #2892119 - Add release adds a release with track 9 of length 8:21 and track 10 was a data track (a trailing data session after audio session, not a track, actually).

But in reality, real track 9 length is: 8:21 − 2:30 = 5:51

You can never know if this or that Free DB import or Add Disc ID edit is right or wrong.
But it’s good to know, when you compare with other releases or recordings, that a length difference of 2:30 on a last audio track should ring a bell: Edit #50657905 - Merge releases


I think I used to have a pretty detailed comment (in old forum topic that contained loads of examples, I guess) about that often seen in old added releases bug.
I am waiting for the return of this forum because I have not saved my topics and I have dozens of useful links to it (not saved by web archive).
So I create this topic here because it’s not very clever to always link to dead forum link in my edit notes.

9 Likes

My recent experience with this issue has been a 150 frame (2 second) longer last track. Of my 40+ Enhanced CD’s there are at 39 discid entries in the database that have the correct discid and the 2 second longer last track discid.

1 Like

It could as well be a home made burned AUDIO only copy of the release, though.

Interesting thought, could be, but for ±39 different artist/release?? My bad discids based on the faulty TOC (150 frames longer) from my old XP media library program matched 29 discids, the correct discids matched more than those 29. Still no way to prove one way or another. All my stuff is correct now and I removed the 10 bad discids I added. In your case you were talking about minutes (the entire 11400 frame gap being added), I was talking about 2 seconds being added so my issue probably does little harm and is in the norm of a different pressing.

1 Like

Just to gather more information on this bug from the various MB archives:

Disc ID differs on different platforms because of 2 second leadout in/exclusion — trac #536archived!

It’s a ticket about something else but it largely mentions the 2:30 bug:

dupuy: « The other problem is that older versions of the Tagger (and many FreeDB clients) don’t handle multisession discs correctly, and this leads to the tailing data track being included, and the penultimate track being 2:30 too long. Presumably this is the reason for the difference between cdlookup and cdda2wav. »

And even points to a then generated BadWin32CDTOCs report — archived! — as well as the same lost forum summary 2:30 bug topic.

An automatic fix of those was mentioned but I don’t think it has been coded as I still come across some 2:30 bugged tracklists.

2 Likes

It’s very confusing.
Today I notice that for an Enhanced CD EXPO EXPO, the correct Disc ID (which last track duration matches actual sound) is the one that is 2:30 longer than the incorrect Disc ID:thinking:

I’m posting an update here to tell that I came across a super informed article about this bug.

In fact it’s a 2:32 bug (11400 frames, 152 seconds), not a 2:30 bug!

I have updated our FreeDB wiki page, in rather bad English, so any review and fix to my changes, there, would be welcome.

4 Likes