Multiple Recordings
How many pieces of music do you have multiple recordings of?
In pop/rock, it's mostly limited for me to They Might Be Giants. With 475 TMBG tracks in my library including multiple compilation albums (which probably shouldn't count) that's no surprise. Some of the double-ups are studio versus live versions.
There's a lot of doubling-up on Jazz standards, Mozart and Bach (helped by the fact I have both the complete works of Mozart and Bach as well as many individual recordings).
But what about more than 5 recordings? I can think of only three off the top of my head: two of them are Charlie Parker's Yardbird Suite (my favourite Jazz piece) and Bach's Goldberg Variations (including 3 recordings just of Glenn Gould if you count the Zenph reconstructions).
The third is in a totally different category as I actually have over 10 recordings of it.
It's Prokofiev's D minor Toccata.
Comments (2)
msh210 on July 18, 2007:
I've been thinking Google should have a music search (for its site and for its Desktop application which searches .mp3's, .wav's, et al. for (1) lyrics and (2) tunes (MIDIs too). (Having it search your own computer for duplicates is one application, which is how it ties in to your post, JT.) Obvious problems:
(1) Lyrics. Extracting lyrics from these music files requires really good speech-recognition software -- so good, in fact, that I doubt it exists yet.
(2) Tunes. This should be easier to search on -- especially for MIDIs. The problems are (a) the user interface and (b) the search-term-matching algorithm (or whatever it's called). (a) How should users specify a tune? You'd need a complicated interface for typing in a tune, possibly by clicking an on-screen keyboard. (b) How do you know whether two tunes are sufficiently similar to match? You need to account for the fact that the tempo might be different; that the user may have typed the wrong notes (either by having all his notes one fifth, say, higher than appropriate, or by getting one or two or n of the typed notes wrong); that the user may have skipped notes (so the algorithm would have to match not only exact matches but also cases where the user's notes are the file's _main_ notes (for some definition of 'main')); and doubtless other issues.
The issues in (2) don't seem insurmountable (to me, not that I know the first thing about how to implement such a thing); the issues in (1) seem much harder unless and until we have great speech-recognition software; once we do, (1) is also not hard.
Incidentally, to answer your question, I don't think I have any duplicates; but otoh I have very few music files. I do have lots of JPEGs (photos), and I'm sure I have some duplicates.
Add a Comment
Last Modified: July 17, 2007
Author: James Tauber
Matt on July 18, 2007:
I have the dubious distinction of possessing over ten versions of "Walking on Sunshine", "Hey Ya!", "Tainted Love", "Cruising for Burgers" and a couple others I can't quite remember of the moment... Collecting covers is fun.