Can I use Shazam

Tags: #<Tag:0x00007f4d4e8127f0> #<Tag:0x00007f4d4e8126d8>

When MB Picard’s Lookup, Scan and Lookup in Browser all fail to find a match for an audio file, I play the file on my laptop and tap to Shazam on my iPhone.

This finds a match in the Shazam app. From what I am seeing, Shazam is a song-centered database, not release-centered like MusicBrainz. How do I proceed to match the Shazam-identified audio file in Picard to the MB database? I can share the Shazam match from the iPhone app to my laptop which gives me this link:

https://www.shazam.com/track/134605919/brandenburg-concerto-no-3-in-g-major-bwv-1048-iii-allegro?referrer=sharemore

Or is Shazam not the best 4th option (after Picard’s Lookup, Scan & Lookup in Browser functions)? Is there a better 4th option when stumped on an audio file?

2 Likes

Actually MB Picard Scan does find a match:

Which conflicts with Shazam’s identification. So which is correct?

Shazam says - The Violinists Ensemble Of Bolshoi Theatre & Boris Haykin
Picard scan says - Orchester der Wiener Staatsoper, Hermann Scherchen

They both agree on what’s being played:

Shazam - Brandenburg Concerto No. 3 In G Major, BWV 1048: III. Allegro
MB Picard scan - Brandenburg Concerto No. 3 In G Major, BWV 1048: III. Allegro

But on who’s playing it, Picard scan and Shazam differ.

How to proceed?

The icon for the match in Picard is yellow, which indicated a lower quality match than green would. So is Shazam’s match better quality? Shazam doesn’t offer green or yellow or any indication of the quality of its match.

When you have a recording that came from a release, the AcoustID fingerprinting system that Picard uses is probably going to be more reliable. It’s able to analyse a file with greater detail and fidelity than Shazam can manage with a noisy acoustic signal.

Of course, that doesn’t necessarily mean that Picard’s result is correct, either. Compare the file metadata in the left and right columns in Picard. The yellow could be a result of dissimilarities in file tags, or in the length of the recordings. If you find that both lengths are definitely correct, yet they are different, then the release is not a match. By default, the fingerprinting algorithm only examines the first two minutes of the recording. But, if the yellow is just because of gaps or minor differences in your file tags, there is a good chance that you identified the recording.

3 Likes

Some options if you are going to investigate further:

If you have other files of parts of Concerto No. 3 in G major, BWV 1048 in your files then try Shazam and AcousticID on them.

Get the metadata (tracklength specifically) for Orchester der Wiener Staatsoper, Hermann Scherchen playing Concerto No. 3 in G major, BWV 1048: III and compare that length to the length of your Track. (Ideally you’d get millisecond accurate times from MB Disc ID or something similar and compare them.)
Here is one matching Recording by Haykin:


That Recording is from

If you have other files of tracks from that Release that would create an identifying context for the mystery track.

Alternatively if you have other tracks from either of the MCA Releases of Scherchen’s Recording then you’d have a different context.