Some more important requirements of an automatic music tag editor

Some more important requirements of a rules based music tag editor

Since this post I realised I'd missed some other important requirements for a fully automatic music tagger

1. There needs to be an audit trail
If you are trusting some of the most important data on your computer then you really want to know what it is getting up to.
 
2. There needs to be a way to roll back changes
However good automated matching is there should always be the possibility to undo the changes
it makes.

3. There needs to be a way to remove duplicates songs
As your song collection gets organised you'll probably find you have duplicates, and finding and removing duplicates is something that automated taggers can potentially be very good at.

4. A way of continuously monitoring and modifying your music collection.
Once you are happy with your automated setup wouldn't it be nice to not even have to think about it anymore.

and this is how SongKong meets these requirements

There needs to be an audit trail

SongKong creates a comprehensive report with details of exactly what has been matched to the Musicbrainz and Discogs databases and also exactly what changes have been made to your files.

There needs to be a way to roll back changes

Every time SongKong makes changes to a file the changes are stored in a database, and because it is in a database those changes are not lost when you close SongKong or restart your computer. If at a later date you decide that you do not like the changes that SongKong has made you can use Undo Changes to change the files back to how they were before changes made by SongKong. This undo facility works even if the files have been moved or renamed.
 
There needs to be a way to remove duplicates songs

SongKong lets you find duplicates, and when a duplicate is found decide the criteria for which if the duplicates to be deleted. But what is a duplicate, luckily because SongKong stores ids when matching songs it can accurately determine when a song really is a duplicate. SongKong lets you choose any combination of Acoustic Id, Song Id and Album Id to let you decide what is a duplicate. For example if the same song appears on two different albums  then you may consider these as different songs or you may consider them the same song.

A way of continuously monitoring and modifying your music collection 

Once you have SongKong configured to your liking a good way for working with new music is simply to setup a new folder that you dump new music into and have SongKong detect this and do its thing, this is easy done using the Watch Folder option.

Rules Based Music Tagging

With conventional music tagging there is lots of manual editing or semi-automatic tagging but SongKong is a rules based tagger. The basic idea is that rules define how you want your music collection to be organized, then the rules can be applied to your whole music collection without any manual editing required giving a totally consistent music collection with the minimum effort on your part.  You can change the rules and then reapply to the whole collection to maintain a consistent music collection.

There are three main aspects to incorporating rules based tagging, lets summarize the perfect system:

1. It would be possible to correctly identify every file in your music
2. Once a song is identified the database it has been matched to would contain every desired attribute of the song that you are interested.
3. It would be possible to define rules to extract and apply the data in any way required.

And this is SongKongs implementation:

1. SongKong generates Acoustic fingerprints for each song and this can be looked up in the Acoustid database to identify the song, working in a similar way to Shazam. Acoustid currently contains fingerprints for 16 million songs allowing the majority of your songs to be identified, but Acoustid  can only identify the song not necessarily the album. However by combining this with comparing existing meta-data in your songs we can match to the album as well in the MusicBrainz database. MusicBrainz provides high quality detailed data onmore than 1 million albums.

You can influence the importance of metadata such as by specifying a preference of matching albums form certain countries of a preferred format such as CD or Vinyl.
 
So MusicBrainz/Acoustid provides good coverage for most music collections. But we also use the Discogs database as an alternative source.

Even so, we cannot guarantee 100% matching, but testing shows that in the majority of cases approximately 90% of a collection can be matched to MusicBrainz or Discogs.

It has to be said both of these databases do have better coverage for Pop/Rock/Electronic western music than Classical and World Music. But there are many projects ongoing to fill these gaps, for example the Music Technology Group at the Universistat Pompea Fabra is working with MusicBrainz to add Indian Raag Music.

The good news is that MusicBrainz open approach is fast becoming the de-facto standard database of music information.

2.  Some Music databases contain only the most basic information such as artist, album, title and year of release. But the MusicBrainz database consists of an incredibly rich semantic model, and allows new relationships to be defined between entities.

When SongKong matches to a song in MusicBrainz it is guaranteed to find at least 19 fields (such as artist, album ecetera) and often an additional 35 fields. This includes high quality artwork, usually at a resolution of at least 600 x 600 pixels.

These fields also includes some MusicBrainz Ids, this means that  your songs are compatible with other MusicBrainz enabled applications. It also means you can always look up the original source of the data at any time in the future.

3. SongKong is intended to be easy to use, but there is no one right way to organize your data. Everybody has different requirements based on their personal preferences and how and where they are going to play their music.

Here are a few scenarios handled by SongKong:

You are a DJ only interested in individual songs and has no interest in the album information. SongKong lets you specify exactly how files are named and stored, you can use any piece of metadata for deriving your filename ,and can use powerful Javascript expressions to manipulate the names.

Your music is already stored in iTunes. SongKong can work with iTunes automatically, updating the information in iTunes for songs that are already under iTunes control and adding songs that are not iTunes control to iTunes. 

You like to process songs in a pipeline moving them from unmatched to matched location. SongKong allows you to move files as they are matched to a new location, it also allows files that it failed to match to moved as well.

I hope this post helps you understand the basics of rule based tagging. Jaikoz also incorporates rule  based tagging , but also provides manual editing and semi-automated tagging.