anyone have a favorite shell program to write AcoustID and Musicbrainz tags

[pfarrell]

I find Picard to be fairly easy to use and works well for 90% of my tagging. But it can be mislead if the FLAC files were accidentally tagged with ID3v2.3 tags But I find it frustrating to use when I know that the tagging is wrong. I'll do a "open folder" and if Picard has seen the folder before, it doesn't display the files. Sometimes it shows what it thinks is the album, but when the tagging is wrong, that can be a random string.

I've tried to install puddletag, well I have installed it, but I can't get it to work because it is whining about X-windows.

I know, the world is all GUI and smartphones. But I'm old fashioned.

Anyone have any tool suggestions?
linux preferred, but WIndows cmd/powershell OK too

[RobbH]

I asked a similar question more than two years ago and got some helpful suggestions... IF you sometimes use a gui on other machines.

https://forums.slimdevices.com/forum/user-forums/ripping-encoding-transcoding-tagging/109816-command-line-tag-editor#post109816

[MrC]

Pat,

You can have Picard perform acoustic fingerprint based look-ups to identify tracks, and then group them into the correct album. That may be all that you need.

Many are happy with mp3tag; however, it is essentially only a tagger, and not a higher-level cataloging / curation tool. But it's free.

If you want much more control, consistency and the ability to fully catalog/curate your media, I'll suggest using JRiver's Media Center along with my MCUtils command line tools. Both run on macOS, Windows and Linux. To get a tutorial-like overview of the suite, see the Docs/User_Guide.pdf at that link. I've been managing my assets using these tools for nearly a decade now, and I'm still actively developing the toolkit. MC + MCUtils for management, LMS for playback. I used to also use MC for playback, but no longer have that requirement as we have a whole house audio system, audio fed by LMS. In case you have that need, MC's playback is outstanding,

[pfarrell]

You can have Picard perform acoustic fingerprint based look-ups to identify tracks, and then group them into the correct album. That may be all that you need.

Yes, that is most of it. But I can not convince Picard to add (write out) the AcoustID fingerprint into the meta data. I have "submit AcoustID fingerprints" so I trust they are going
off somewhere in Musicbrainz's cloud.

Seems so simple, since Picard is already writing out lots of metadata....

[pfarrell]

I asked a similar question more than two years ago and got some helpful suggestions... IF you sometimes use a gui on other machines.

Thanks, I replied to your old post, offering to write something if you still need it. I'm writing assorted utilities for linux, what you need is nearly trivial. I often prefer shell/command line programs, as you can write programs to write the commands to handle thousands of songs.

[MrC]

Yes, that is most of it. But I can not convince Picard to add (write out) the AcoustID fingerprint into the meta data. I have "submit AcoustID fingerprints" so I trust they are going
off somewhere in Musicbrainz's cloud.

Seems so simple, since Picard is already writing out lots of metadata....

The fingerprint data you get in Picard is indirect - it's via reference to an MB UUID value. This (via Lookup in Browser) can take you to the page on AcoustID:

https://acoustid.org/track/92a763c0-bf96-4664-ac35-4f552e80af20

And verified using MC:



Be sure you have this setting enabled:



Finally, you can review:

https://picard-docs.musicbrainz.org/en/variables/tags_basic.html

[pfarrell]

I made sure that "Save AcoustID" was checked, restarted Picard and still don't see it in the windows.
Here is a screenshot.

[MrC]

I made sure that "Save AcoustID" was checked, restarted Picard and still don't see it in the windows.
Here is a screenshot.

Once Picard has Release MBID's or Track MBID's, it no longer allows you to click the Scan button, as acoustic fingerprints become superfluous. The IDs are present, so there's no need to perform that form of lookup. Picard can more efficiently lookup info using the MBIDs.

You have to remove the MBID tags, save the file tags, remove the entry from the right side of interface, then (re-)Load the file info so that the tracks appear in the Clustered/Unclustered Files area. Then, you can Scan to generate the IDs, and save them.

While you can generate the fingerprints of files at any time in Picard, that alone is insufficient as fingerprints will hit potentially many Release IDs. So it is necessary to allow Picard to first Scan, then select the correct Release ID, then your tag Save will have those tags available.

If all you care about is the actual fingerprint, you can get it more efficiently this way:

$ /Applications/MusicBrainz\ Picard.app/Contents/MacOS/fpcalc ~/Desktop/Temp/temp\ music/01\ Minuano\ \(Six\ Eight\).flac
DURATION=567
FINGERPRINT=AQADtEkUSYmSD_2RL3iO4w-ao8ePeCc0Ho3aCtmuo5mP_EQzJsfbBNKcHVaPqzp-dHqKMDmuBGEfNA9x7...

[pfarrell]

If all you care about is the actual fingerprint, you can get it more efficiently this way:]

I don't want to see it, I want to have it written into the meta data.
Doesn't seem like anyone else cares about that.

[MrC]

I don't want to see it, I want to have it written into the meta data.
Doesn't seem like anyone else cares about that.

I get that. I was just trying to explain that Picard is the wrong tool for this, once your tags already have the MBID values, since those are more accurate.

And since you are already writing a command line tool, just call fpcalc to return the value, which you can write.

I used to do exactly this sort of thing with a different fingerprinting tool, but I obsoleted it when the fingerprinting server was removed from public use, and made proprietary. The server returned several musicality values, so was useful. But since then, MB has grown substantially, and MBIDs are far more useful than a fingerprint. So I never bother updating the tool to use fpcalc, esp. since it does not provide any musicality data.

[pfarrell]

I was just trying to explain that Picard is the wrong tool for this, once your tags already have the MBID values, since those are more accurate.

And since you are already writing a command line tool, just call fpcalc to return the value, which you can write.

Thanks, yeah, its not the tool I'm looking for. When I manually run fpcalc, on a typical file, I get about 3700 characters of output. The AcoustID fingerprint IDs that I've seen Picard display are much shorter.

I don't have any at hand, but I think they are more like 64 characters in some base64 or similar format. And they are broken up into fields or separated by occasional hyphens.
I don't see any mention of this shorter format in the man pages for fpcalc.

[MrC]

Picard can write both of AcoustID, and AcoustD Fingerprint. The former is a UUID that is used to construct a track URL on the acoustid.org site (or if you right-click in Picard, you can Lookup in Browser). The latter is the actual fingerprint returned from fpcalc. That fingerprint is used to query the acoustid.org server, which the can return the UUID shown in the AcoustID tag.

And that AcoustID UUID value is what Picard uses to lookup the various MBID values for the track or set of tracks. But once you have those, the AcoustID and AcoustID Fingerprint tags become rather superfluous, as I mentioned.

Below is the process, visually, when you perform a Picard Scan on a track that has no MBIDs already. I show the linkages regarding how the data is used to construct the standardized URLs used by acoustid.org and musicbrainz.org:



The sole purpose of the AcoustID fingerprint is to return the AcoustID UUID. And its sole purpose is to present in Picard the possible albums to which the track belongs. Once you've selected the Release, Picard no longer needs the AcoustID UUID value, nor of course will it re-run fpcalc to generate the long AcoustID fingerprint value.

The MBID values are key. The AcoustID information is simply a means to an end - that is, selecting the correct MBID values.

Hopefully this is helpful to you (or anyone else who is interested in how this works).

[pfarrell]

Hopefully this is helpful to you (or anyone else who is interested in how this works).

Thanks. I've noticed that using the "goto browser" button on Picard will sometimes show many albums for a specific AcoustID. I would expect this if a single digital "master" of a song is sent to a bunch of producers for "greatest hits" and "rock of the 1970s" type collections.

[MrC]

Correct. A fingerprint may match multiple tracks. There is no guarantee of a 1 to 1 mapping between a fingerprint and a track. That's where the user is required to select the correct one.