This proposal implements the changes to recordings discussed in the three recordings meetings held during January 2013. The meetings were open to all of the MusicBrainz community, and issues were discussed thoroughly, so any criticism of this proposal should hopefully be focused on the wording, and not the principles.
This proposal replaces the current Recording Style Guideline with the new revision in the section below.
See the title guidelines for how to enter the recording's title.
The recording artist should be the main artist involved in the creation of the recording. Usually, this will be the performer of the song, but it could also be the artist that remixed or recorded the recording.
A disambiguation comment should be used to identify distinguishing features of a recording where two or more recordings share the same name.
In the specific case of live recordings, the naming format from live bootlegs should be used.
- Train in Vain (live, 1998-12-14: Telewest Arena, Newcastle, UK)
- Wake Up (live, Los Angeles, CA, USA)
- The Dance (live, 2002)
- Candle in the Wind (single edit) (live)
It can be difficult to know which recording to use on each track of a release, but generally, the most obvious recording will be the correct one.
As explained in Recording, a recording is a unique mix. In most cases, a track will be the original mix of a performance of a song. However, there are some important cases to consider where this is not true - these are discussed in the following section.
Different performances of the same work should always be given separate recordings, no matter how similar they may sound. This applies to both studio performances and live performances.
Different Recordings of the Same Performance
This section mostly applies to bootleg recordings. Different audio recordings of the same performance will always result in different audio stems - when the stems are mixed, the resulting mixes will always be different. A new MusicBrainz recording should be created for each mix of a performance using different audio recordings.
Remixes and Remasters
Firstly, an existing mix can be taken and used as the basis for an entirely new mix. This new mix will either be a remaster or a remix of the original song. Remixes and to some extent remasters, should be represented by new recordings.
A remix involves mixing the performance for a different purpose - for example, broadcast on public radio or playing in clubs. A named remix should always be given its own recording in the database. Remixes which aren't explicitly named differently to the original mix should be given a new recording, if there is evidence that supports it being a different mix.
"Remaster" is a bit of a misnomer. A remaster typically involves "refreshing" the audio of the original stems used in mixing, and correcting any errors in mixing in earlier copies. Remasters should only be given new recordings if there is some evidence that new mixing has been done to produce the remaster.
In addition to the above guidelines, the following rules should always be followed when considering merging two recordings.
Recordings with Different Durations
Recordings of different durations can be merged, as long as there is no evidence to suggest that mixing has been done to change the lengths. Volume fades at the beginning or end of a song aren't considered "mixing" unless they cause the structure of the song to change.
For example, Recording A which slowly fades out on the last chorus of a performance can be considered the same as Recording B which has the last chorus at a constant volume. On the other hand, if Recording C of the same performance faded out during the verse preceding the chorus, and cut out before the chorus was heard, that would be considered a separate recording to Recording B.
Recordings with Conflicting Relationships
Generally, don't merge recordings if they have conflicting relationships. However, if you're certain that two recordings are the same and relationships are wrong, merge the recordings and correct the relationships.