Installation

Picard is a cross platform tagger that is written in Python and available for MacOS, Windows and Linux. Thankfully it comes with a proper installer so can be easily downloaded and installed for any of these platforms, because all versions share a common code base the functionality on all platforms is the same.

User Interface

The User Interface (UI) does vary to fit the platform it is running on is, for example it supports macoS Dark Mode. But we did not find the UI not very intuitive and it is not obvious how to use it correctly. To use the main feature of Picard (automated tagging) the user has to make decisions about clustering, fingerprinting and scanning which require non-trivial technical understanding.

Song Identification

Picard is the Music Tagger provided by MusicBrainz so unsuprisingly it properly understands the MusicBrainz database and can tag your music files, it is a decent tagger but there are some difficulties with tagging compared to SongKong

Picard can only find matches if the albums are in MusicBrainz. We agree that MusicBrainz is the best database and it is also the primary database used by SongKong but it is not the largest database so only using MusicBrainz is going to limit how many of our albums can be matched. SongKong also uses the much larger Discogs database and the Acoustid database. The number of songs in the Acoustid database is much larger than the MusicBrainz database, whilst Picard does use Acoustid to fingerprint songs it will not identify them if they are not matched to a MusicBrainz song, and only one quarter of Acoustids match to a MusicBrainz song.

The song/album identification is not fully automated, the customer is expected to check all results before saving, and this is not reasonable when have many albums to match. Within Advanced:Matching:Thresholds we can modify the thresholds for matching songs and albums but without an intrinsic understanding of how the matching algorithm works it makes no sense to be able to do this. Instead Picard should have a robust algorithm that we can use with confidence knowing it will rarely do incorrect identification.

We found that trying to tag more than 1000 songs in one go just caused it to hang, and although Picard can be used to tag multiple albums in one go they generally recommend working one album at a time which is much slower

Metadata Formatting

From our testing we measured that Picard can add about sixty metadata fields which is more than most taggers, but SongKong can add over twice that (128) so where do the differences lie?

Picard does not do alot for Classical Music so all performers are added to the Performer field, and it can also add to the Composer and Conductor fields, but it does not add to fields such as Orchestra, Choir and Ensemble. It only adds the standard sort fields so it adds to Artist Sort but not Conductor Sort or Performer Sort. It does not add acoustic information such as Bpm or Mood. SongKong can categorize albums with useful boolean fields such as Is Classical, Is Greatest Hits and Is Soundtrack but Picard does not.

Picard does not provide a log of report of the changes it has made so it is difficult to keep track of what has changed over time, and there is no option to rollback the changes after they have been saved.

Tag Flexibility

The way Picard adds tag metadata added to our files is very tied to the metadata as it is stored by the MusicBrainz database, so sometimes the data is added in way we consider that is not very helpful.

For example MusicBrainz support a wide variety of media types and Picard can store the media in the media field, however the list is very large an unwieldy containing many record types such as 7" Vinyl, 7"; Shellac, 10" Vinyl, 10" Shellac, 12" Vinyl, 12" Shellac. In contrast within SongKong we simplify the whole list to only Vinyl, CD, Cassette and Digital Media, this is actually more useful for the majority of users.

Sometimes classical releases are stored in MusicBrainz using the Classical Style Guidelines and sometimes they are not. When using CSG unlike the normal guidelines the track artist contains the composer, and performers are stored in the recording artist. But Picard doesnt do anything special and stores whatever is stored as the track artist as the track artist, this is incorrect for Classical releases. In contrast SongKong identifies classical releases and does special processing to ensure the artist values are allocated to the correct tag fields.

Manual Editing

Direct manual editing is limited to simple edit of the fields of the selected song(s), but that is understandable because Picard main aim is matching albums to MusicBrainz rather than manukl editing.

Picard does have Tagger Script and this allows tags to store custom values using a scripting language. For example the Release Type plugin can be used to add the release types to the Album field. It looks like we can do alot with tagger script it is a powerful feature that SongKong does not currently have. The difficulty with this approach is that using predefined tagger script is to risk using something not maintained by the Picard developers themselves, and to use and write tagger script ourselves requires some basic to advanced coding skills, and this could allow users to create scripts that may not work as expedcted.

Rename Files

Picard has a comprehensive File Renaming scripting using similar syntax to Tagger Script

Delete Duplicates

Picard does not have a Delete Duplicates task

Support

Picard has a user forum and public bug tracker and there is plenty of help online.

Summary

Picard is reasonably good at extracting data from MusicBrainz into your music files, but can only work with a limited number of files at a time and is rather ecoteric and clunky. It is not a general purpose music tagger.