Hi,
i am trying to obtain rhythm strength and rhythm regularity by calculating the
onsets followed by autocorrelation.
I was hoping I could take a temporal approach (10 second mid-term frames) to
extracting and averaging the onset peaks from the envelope. I was also hoping
I could autocorrelate each 10 second frame, perform a peak detection and
average the autocorrelation peak's amplitude and derive at 1 statistic for the
whole audio track for average onsets and average correlation peak amplitudes.
The only way I can see this working so far is by taking the onset detection
(mironsets) of the whole audio track (~3 mins), then performing
autocorrelation, then getting the peaks using mirpeaks.
Thanks.