Re: Software for chord identification (robert gjerdingen )


Subject: Re: Software for chord identification
From:    robert gjerdingen  <r-gjerdingen(at)NORTHWESTERN.EDU>
Date:    Wed, 7 Sep 2005 18:04:35 -0500

On Sep 7, 2005, at 5:21 PM, Eric Jacobs wrote: > If I know what chord is being played (call this "actual chord") - and I > do know what chord should be played (call this "ideal chord") - I can > adjust the recording speed as required to reproduce the ideal chord. Eric, I agree with Leon that "chord" may not be the best entry point to determine original pitch and therefore turntable speed. Since chords are aggregates of pitches, and since knowing the "ideal" chord means knowing the ideal pitches, and since early recordings strongly foreground the vocalist or soloist, it would probably be sufficient to isolate a known "ideal" pitch and run it through a fundamental-pitch-extraction algorithm. While it might be possible to have a high degree of confidence in the ideal pitch of early classical instrumental recordings, confidence would be much lower in popular or classical vocal recordings. In both realms pitch was often adjusted up or down for the comfort (or vantity) of the singer. Nonetheless, if you really do know the ideal pitch, then the degree of error in pitch determination would probably be less than for other measures (e.g., vibrato rate). On a different tack, if your recordings are of dance music, you might be able to use high-confidence recordings (e.g., movie soundtracks) to estimate what the Italians once called the _tempo jiusto_, or the "right tempo" for a given genre of dance (e.g., foxtrot, tango, etc.). Currently the very best software for comparing the actual pitch to the ideal pitch is known as a musician :) Bob Gjerdingen Northwestern Univ.


This message came from the mail archive
http://www.auditory.org/postings/2005/
maintained by:
DAn Ellis <dpwe@ee.columbia.edu>
Electrical Engineering Dept., Columbia University