SACD types and DiscID

Tags: #<Tag:0x00007fe31511a738>

Medium type “SACD” should never allow DiscID, cf. - it has no CD layer like a Hybrid SACD does.

An edge case is medium type “SHM-SACD” which is basically a SACD but there might be some releases with CD layer - I need to investigate more.

Wouldn’t it be SACD (SHM-CD layer) and SACD (SACD layer)?

No. Please read again by daylight :slight_smile:

Actually it seems I am wrong. :slight_smile:
I don’t know anything about it, it was just my understanding of the new format name.

This link above contains some probably more knowledgeable SHM-SACD discussions.

It seems I am wrong as well… I was under the impression that SHM-SACD offered a CD layer but going through my discs I can’t find an example for that at the moment… :blush:


I’m not at all knowledgeable about SACD / DiscIDs, so I’m using this opportunity to ask:
I was looking recently through the DB and found ~50 discds above 80 minutes in time, usually 10 to 50 hours discs. I guess those come from DVDs or SACDs. They are attached to CD, SACD, Hybrid SACD, and Hybrid SACD (CD layer) (cf examples at the end of this message)

Does that make sense to have those discIDs on MusicBrainz, since they don’t represent the real length of musical tracks?
Should those be restricted to SACD only? Or automatically rejected?
I don’t think we have anything clear in the documentation / guidelines about this case.

CD │
Hybrid SACD │
Hybrid SACD (CD layer) │

CD release with both kinds of discIDs:
CD with 43hours discID applied:

(SQL query)


Disc IDs represent the TOC of a compact disc. So they don’t make sense for other media types.

I suppose it’s possible also that a CD copy protection scheme could have busted up the TOC in this manner, for example perhaps a multisession disc could be made with only the first session having a valid TOC (with the idea being that your CD player does not implement orange book multisession discs but your PC does and gets confused).

1 Like

It’s quite possible these errant discids were submitted with old versions of foo_musicbrainz, a component for foobar2000. TOCs were calculated based on the number of samples per track and not reading from CDs.

With older versions, for a selection of tracks to “qualify” as CD, the length in samples had to be divisible by 588 to leave a whole number which is fine if the samplerate is 44100 (44100 / 588 = 75 sectors)

With the samplerate of SACDs being 64x44100, it would still be divisible by 588 but the each track length would be 64 times longer than the reality.

But any like any TOC submission, it has be done through a browser and any editor using this and not even glancing at the generated track times is an idiot and I think only they are to blame for these entries!!! These entries absolutely should be removed.

edit: harsh comment about editors not noticing abnormal times removed. It seems the site doesn’t provide a nice preview before submitting. Apologies. :smiley:

FWIW, the component was originally written in 2009 and abandoned in 2012 by the original developer. I made some minor tweaks in 2015 and then gave it a major rewrite last year and this bug was fixed by me as it was also reported/discussed in this thread…

I no longer maintain it.


To be fair to these editors the disc ID submission user interface is simply awful and nowhere does it display the track times of the disc ID you are about to submit nor does it display the existing track times (if any) of the release you are about to attach it to. The only feedback you get is the total number of tracks.

You can only get a nice display of the disc ID compared with current track times after it is attached to a release, by beginning to enter a “Set Tracklist Duration” edit (for example with one of these bogus disc IDs)

I always double check immediately after disc ID submission (usually by submitting an update track durations edit) because I have personally made disc ID attachment mistakes on multiple occasions. It would be really nice if all that information was displayed before the disc ID is attached…


OK, I take the “idiot editor” comment back. Also, I guess it could be argued some sort of server validation for maximum lengths should also be performed!


But we can remove them all, those wrong Disc IDs.


@marc2k3: thanks or the link to the old thread, I completely missed that one.

So the consensus seems to be “all discIDs above the physical limit of the CD can/should be removed, since they had to be computed from buggy tools/corrupted discs”, right?

I will create a report for discids > 90 minutes to help the cleanup.

Right now I see in the DB:
name │ count
CD │ 119
CD-R │ 1
DVD-Video │ 1
Enhanced CD │ 1
HDCD │ 1
HQCD │ 2
Hybrid SACD │ 52
Hybrid SACD (CD layer) │ 9
SACD │ 26
SHM-CD │ 1


That would be great. So far I used ReleasesConflictingDiscIDs (for subscribed entities) to find them by hand :wink:


Yep, 90 minutes seems fair. I vaguely remember “overburning” being a thing where some home-made CD-Rs might be slightly over 80 minutes.

I was about to say anything over 80 minutes wouldn’t be commercially available but then I had a quick google…


I submitted a report for inclusion in the official list, if somebody wants to start working there’s a static version on github :slight_smile:

Could the general “Hybrid SACD” be blocked so that DiscID can only be added to "Hybrid SACD (CD layer) explicitly? In that case I can safely submit some SACD layer track times when adding a general release 1-medium-style without being in danger of losing those to a DiscID edit.

Why don’t you add the CD layer at the same time?
Isn’t the CD layer easier to add to MB?

1 Like

Wouldn’t that create duplicate recordings?

Not if you do like this:

1 Like

Hm… I tried that before and it doesn’t seem to bring up a medium that is just being added, only existing ones. I’ll look again.