Selection of preferred disc id among discs of same content

I own Journey to the Isbin, performed by Sharon Isbin. The album was purchased individually and the artwork in the physical media appears as shown.

Long ago, I ripped using Exact Audio Copy, which attached desired metadata to the resulting whole-album FLAC file and cue sheet file.

Recently, I attempted to update the tags from MusicBrainz (using the flactag utility).

The discid computed from the cue sheet corresponds to a particular release in a boxed set, very different from the product I purchased.

The correct release in the database corresponds to a different discid.

Why is the MusicBrainz data apparently inconsistent with the purchased product, and how may I force flactag or other applications to apply the more appealing and accurate representation?

The cue sheet is as follows:

CATALOG 0630198990229
FILE "Sharon Isbin/Journey To The Amazon/Sharon Isbin - Journey To The Amazon.flac" FLAC
    ISRC DEA809701100
    INDEX 01 00:00:00
    ISRC DEA809701200
    INDEX 00 02:09:32
    INDEX 01 02:11:05
    ISRC DEA809701300
    INDEX 00 05:43:27
    INDEX 01 05:46:02
    ISRC DEA809701400
    INDEX 00 09:43:65
    INDEX 01 09:47:15
    ISRC DEA809701500
    INDEX 00 13:39:30
    INDEX 01 13:44:45
    ISRC DEA809701600
    INDEX 00 19:12:65
    INDEX 01 19:15:65
    ISRC DEA809701700
    INDEX 00 23:41:02
    INDEX 01 23:42:15
    ISRC DEA809701800
    INDEX 00 25:09:45
    INDEX 01 25:11:02
    ISRC DEA809701900
    INDEX 00 27:33:35
    INDEX 01 27:37:42
    ISRC DEA809702000
    INDEX 00 29:53:17
    INDEX 01 29:55:62
    ISRC DEA809702100
    INDEX 00 32:34:02
    INDEX 01 32:40:15
    ISRC DEA809702200
    INDEX 00 35:26:55
    INDEX 01 35:27:37
    ISRC DEA809702300
    INDEX 00 40:44:50
    INDEX 01 40:46:50
    ISRC DEA809702400
    INDEX 00 44:28:65
    INDEX 01 44:29:67
    ISRC DEA809702500
    INDEX 00 45:56:27
    INDEX 01 45:58:47
    ISRC DEA809702600
    INDEX 00 48:41:25
    INDEX 01 48:44:32
    ISRC DEA809702700
    INDEX 00 50:26:30
    INDEX 01 50:29:10
REM FLAC__lead-in 88200
REM FLAC__lead-out 170 144398100

The TOC from the EAC log is as follows:

     Track |   Start  |  Length  | Start sector | End sector 
        1  |  0:00.00 |  2:11.05 |         0    |     9829   
        2  |  2:11.05 |  3:34.72 |      9830    |    25951   
        3  |  5:46.02 |  4:01.13 |     25952    |    44039   
        4  |  9:47.15 |  3:57.30 |     44040    |    61844   
        5  | 13:44.45 |  5:31.20 |     61845    |    86689   
        6  | 19:15.65 |  4:26.25 |     86690    |   106664   
        7  | 23:42.15 |  1:28.62 |    106665    |   113326   
        8  | 25:11.02 |  2:26.40 |    113327    |   124316   
        9  | 27:37.42 |  2:18.20 |    124317    |   134686   
       10  | 29:55.62 |  2:44.28 |    134687    |   147014   
       11  | 32:40.15 |  2:47.22 |    147015    |   159561   
       12  | 35:27.37 |  5:19.13 |    159562    |   183499   
       13  | 40:46.50 |  3:43.17 |    183500    |   200241   
       14  | 44:29.67 |  1:28.55 |    200242    |   206896   
       15  | 45:58.47 |  2:45.60 |    206897    |   219331   
       16  | 48:44.32 |  1:44.53 |    219332    |   227184   
       17  | 50:29.10 |  4:05.15 |    227185    |   245574   

It’s just that someone added the boxset, but nobody added a release with your disc ID to the database. There is no deeper truth here, but data nobody adds is not available :slight_smile:

The release at has rather limited information. The disc ID is suspicious, each track is exactly 2 seconds longer. This often is an indication that the disc ID was taken from a home burnt CD and not the original.

I’d suggest adding your disc ID to that release, together with missing details you have available like e.g. catalogue number. I don’t know what tool you use for the disc ID generation, but probably it can give a submission URL. Picard also could do it, but only if you have the physical CD at hand.


Before submitting your Disc ID this way, it would be interesting to know if the generated Disc ID is the same with or without Exact Audio Copy Detect Gaps (F4) beforehand and also compare with Picard (or musicbrainz-isrcsubmit tool) generated Disc ID - with the official current libdiscid.
To see if it generates proper Disc ID.

If you can do it, please tell in this topic.
I have EAC, so if I have some time one day I’ll try, but I don’t remember how you generate the cue sheet (preferably without ripping).

1 Like

This observation is useful, and would clarify the confusion that originally prompted the inquiry. Otherwise, what appears to be the same release would be mapped to a different disc layout ever so slightly different from mine.

I’m happy to help, but there is much I don’t know still. I am still learning about all the nuances regarding releases, recordings, albums, and so on, since my relation to the music industry so far has been one of only a very casual consumer. What is a catalogue number, and where do I find it for an album I own? Are any straightforward references available to clarify similar terms through a plain explanation?

I’m not sure I quite understand. I don’t know exactly what “Detect Gaps” is in EAC, but I would imagine it is a critical part of the process of ripping an entire disc to one audio file and then building an associated cue sheet. The definition of multiple indices for each track is essentially a gap, right?

It seems you are driving at the same concern as the other poster, that the ripped information used to generate the existing discid was not generated correctly.

I don’t understand the suggestion about ISRC. I have never used them before, but notice they are shown in the cue sheet I have reproduced.

The clue here is that the last track length is the same as the boxset. I have seen this in legitimate re-releases where the CD had been reproduced to include the already existing silence between the tracks and then added 2 seconds to each track but the last (the last track gets no gap).
In one release I added there were 4 different discid’s (including mine), each ((tracks-1)*2) longer than the next shorter discid. On examination the three longer discid’s all the tracks were 2,4,6 seconds longer than the shortest with the last track length being the same length on all. Mine was more than a minute longer than the shortest which caused me to look into it.

@dashv: So the publisher added more intermediary silence, or just changed to gap layout so that the track lengths changed without affecting playback?

On my release there was about 8 seconds of silence between all but the last track. I can not confirm that for the other discid’s on that release since I do not have them, I just compared the track lengths. I rip full album+cue using EAC and can listen to it with VLC since it supports cuesheet. I loaded it into Audacity and you could see the 8 seconds of silence between tracks.

Is the earlier advice still appropriate, of “adding your disc ID to that release, together with missing details you have available like e.g. catalogue number” (from @outsidecontext)?

There is no edit history for the present discid. Your discid appears to be valid as it is the same as in the 50 CD set, so I would add it. EAC retrieves that CATALOG entry from a subchannel on the CD as well as the ISRC’s. Generally the CATALOG corresponds to the UPC or an “original UPC” it is not the labels catalog# in the database which is generally on the spine of the package. If your package is the one from amazon then the UPC is 706301989920. You should not enter the catalog number unless you can visual see it on the package.

Just to be clear, since I am still learning the concepts, the suggestion to “add” the disc, means to associate the Disc ID (which has already been entered into MusicBrainz) with an existing release (or perhaps with a new release entered as a modification of an existing release)?

It is not uncommon for a release to have multiple discid’s, could be multiple pressing plants. You will also find many cases of different Releases within the same Release Group having the same discid due to different countries, changes in the package, re-release, etc.

Yes, adding the disc id to the proper release, which I think should be as you wrote.

I think directly on the MusicBrainz website there is no direct way to copy an existing disc ID over to another release. It would usually be submitted again from a CD, where one then can attach the disc ID to an existing release or add a new one. But given that the cuesheet gives Disc ID “boJ3N1Tjy8qfELDbefu7uV.P.WY-” - MusicBrainz and this provides all the data it is possible to construct the submission URL. This URL would be used to submit this disc ID:


@outsidecontext: Should I make a new release based on the existing one, since the cover art is different for the boxed set versus as the individual album?

Now I’m confused :slight_smile: Do you have the box set or standalone? Does your release match ?

As I understood so far yours matches the linked one, so this could be used and the disc ID added to that.

Btw., does your release have a barcode? Is it
0630198990229? If yes that could be added as well.

Maybe you could provide a photo / scan if your release’s front and back, then we could all figure out the details.

At some time soon I will find my disc and packaging, which is currently in a box somewhere. However, what I own is essentially the product in the first link of my first post. It is a disc sold standalone with the original distinctive cover art.

I have just started using MusicBrainz and I was surprised to see that the application I was running matched my disc to the title and cover art from a boxed set. I am still trying to understand the design rationale. It seems that MusicBrainz recognizing distinct releases actually complicates the process for the user of obtaining desired tags. In the best case, a user would need to choose among several releases as a source of tags and art. In a slightly worse case, the user would not find the matching release and must choose between using a different one or submitting new information into the system. In an even worse case, the user would believe that a correct match had been made, but be mistaken. Because of the large number of releases, recognizing the difference seems to increase the chances of the user applying inaccurate data.

The CD you own is also part of a huge box set, it is disc 47 in Release “Teldec - 50 CD Collection” by Various Artists - MusicBrainz . Someone who presumably owns this box set has added this to the MusicBrainz database and also submitted disc IDs for all the discs. This is a box set released in 2013, seemingly consisting of a collection of 50 different releases by TELDEC.

The release Journey to the Amazon had been previously released and has a separate entry in MusicBrainz. That release had been added a long time ago, but without much details being given, see . Most notably it has a disc ID attached that is consistently 2 seconds longer than the one you report, which is usually a sign that it was not from the original disc but a home burnt copy (which actually should not be submitted).

Yes, of course. If data is not there somebody has to add it. That’s how a community maintained database works after all.

1 Like

I understand about the community effort. My concern is that distinguishing releases as it is done may represent a much finer level of granularity than serves the interest of most users, because both a correct match is less accessible and an incorrect match is more probable.

It all depends on what you use the data for and about what you care. If you use the data for tagging and deeply care about this being tagged against exactly your 1993 UK version with the odd sticker on it and not the audiowise identical 1994 Europe release, then yes, you need to check in detail. Users who do care about this usually do extensively check the details, e.g. catalog numbers, cover art scans etc. If you don’t care about all of this and just want to tag against a correct track list it probably doesn’t matter.

As you have read in the discussion so far some users even would like to go into much more details, not only distinguishing between what currently MusicBrainz (and probably most consumers) would consider separate releases. but also distinguishing between different production runs and CD pressing facilities as identified by the matrix codes.

But I mean in this case it is a rather clear distinction: One is a massive 50 discs box set, one is a single disc release :man_shrugging: That’s not just minor details of difference.


I agree that various individuals would have different concerns, but I wonder which group is served by such a high level of granularity. A casual user likely will obtain inaccurate information on specific details, even though he would be happy with only the general information, whereas also the fussy user likely be aware of not having found the correct match, and be bothered by it. He may enter the data himself, but if this requirement becomes typical, then the purpose is defeated of sharing data.

It seems to come down the observation that if all the members of a group are trying to hit the same, large target, then most will succeed. If, however, each is trying to hit a separate, small target, most will miss the intended one, which itself is undesirable, but more, many will also hit an unintended one.