Offline Database of only my own music?

Is it possible to make an offline database of my own music using the Musicbrainz system? To create my own personal Musicbrainz site, hosted locally. That only I can access and edit.
I want to be able to see only the Cds and tracks that I own. Be able to compare AcousticIDS and ISICS for the tracks that I own.
I have nearly finished adding all of my music (aprox. 1500 cds) to the database.
I understand that this idea isn’t the original premise of Musicbrainz. Though I am happy to continue to add and edit Musicbrainz for the greater good.




the server code (and install instructions) are on GitHub - metabrainz/musicbrainz-server: The official musicbrainz-server codebase

i’m curious as to your use cases, as nothing comes to mind where a local/trimmed-down instance would be better than using the main collaborative db

1 Like

I would start with the preconfigured MusicBrainz Docker.

1 Like

Maybe I am looking at this the wrong way.
One of the issues I have with the main database is… Looking at a single track on a CD that I own. It will have multiple ISRC codes, multiple Audio fingerprints. How do I know which ISRC & Audio fingerprint is my track?
Adding to that.
If I can’t tell which audio fingerprint is my track, How do I compare multiple tracks to see if they are the same or different?
I thought that if I can create my own database, this would resolve my issues.

It’s probably way outside the scope of anything that can be done, but I’d love to be able to dump the data from one of my collections to a format that could be read by Microsoft Access.

Picard will compute and write the exact audio fingerprint for your file.

However, it will retrieve all ISRCs associated to the recording from the database. The database just do not distinguish between recordings with some minor differences such as remastering. See AFAIK there is no way to distinguish the one ISRC associated to the specific track the recording appears on from the database.

Maintaining your own database won’t solve it unless you don’t merge recordings the same way.

Maybe it would make sense to associate ISRCs directly to tracks instead of recordings in the database?


Copyright relationships on recordings have the same problem as ISRCs in that it is often associated to a specific track rather than the whole recording. If ISRCs are schema-changed to become track-level data instead of recording-level data, I’d do the same for the copyright relationship.

1 Like

I am not keen on having an additional layer for tracks.
IMO we can live with having all ISRC and copyright relationships in one recording when the sound is similar.
Same as we are OK to loose mastering info from recording level.

A couple of examples.

This recording has 2 IRSC codes.
USWB10903611 3:58 2009
USWB19901324 5:18 1983
1:20 difference in length.
With 28 AcoustIDs varying in length from 3:39 to 5:10.

The recording has 2 IRSC codes.
USMO17100531 4:13 1972
USMO17200531 3:54 1972
20sec difference in length.
With 11 AcoustIDs varying in length from 3:47 to 5:33.

Going by this AcoustID database,, which I know has its own set of issues. I would have thought that each example should have its own recording. Then you add the AcoustIDs into the mix.

The Madonna example has 56 combinations of ISRC and AcoustID
The Michael Jackson example has 22 combinations of ISRC and AcoustID

I don’t see the 5:18 track over there.
According to what source can you tell it’s 5:18?

Maybe this ISRC was added wrongly, the original edit has been challengd without answer so you can remove with an edit note saying why: Edit #11980600 - Add ISRCs (pre‐NGS).!/search?isrcCode=USWB19901324&tab=lookup&showReleases=0&start=0&number=10

1 Like

Sometimes wrong ISRC are printed inside CD.
In this case as it’s actually printed on CD, they are no longer wrong IMO as they are de facto attributed physically.

These ISRC probably come from an actual CD.
But as there is no answer from the original editor, you can in doubt remove ISRC from these recordings…