Automatic tempo detection for huge loops libraries ?

[Vince]

Hi,

Out of the grave I have found an old library of drum loops, but none has any tempo indication ... would you know of a way to script Audacity to scan directories and rename the files with the BPM appended in the name ?

Or another method ?

Thanks !

[BertKoor]

Another method? Sort by file size!
If the sampling rate is the same, then size correlates with length in time. With loops: that is tempo.

[Ah_Dziz]

Using a combo of autocorrelation and fft analysis you should be able to extract some very good tempo data. Probably python would be the way to script this for batch processing. I'm not sure of any off the shelf products for this. Many audio editing platforms have robust tempo detection built in somewhere.

[Vince]

Another method? Sort by file size!
If the sampling rate is the same, then size correlates with length in time. With loops: that is tempo.

Different sampling formats, different bits formats, different durations ( 1 bar , 1/2 bar ,etc ) ... the perfect condition to give a headache for someone willing to sort files ...

[Vince]

Using a combo of autocorrelation and fft analysis you should be able to extract some very good tempo data. Probably python would be the way to script this for batch processing. I'm not sure of any off the shelf products for this. Many audio editing platforms have robust tempo detection built in somewhere.

Autocorrelation and FFT ... let mecheck for some Python scripts ...

Thanks !

[Ah_Dziz]

Yep. The tempo is basically an incredibly low frequency bin with an fft. 4/4 kick patterns being super easy to pick apart. Once you are down to those low tempos, autocorrelation can tell you much more about the tempo.