I have a long audio file (12+ hours). I know that there are some unknown small (2 minutes each or thereabout, it varies) repeating (not bit by bit: it is recorded with a mic) chunks in it, repeating not periodically. My question is: is there a completely automated way to find these chunks their offset in the main audio file and the times they occurr - with offsets for each occurrence?

I've heard about autocorrelation for similar problems, but I guess it is useful for small local chunks, not long sequences like mine.

  • 1
    $\begingroup$ Autocorrelation will work even with large files. You may have to roll your own implementation in case the signal does not fit into your computer's memory, but there may be libraries that handle that case already. $\endgroup$ – MBaz Sep 11 '19 at 19:28
  • $\begingroup$ Will it have peaks at the beginning of each repeated chunk? How should I implement the general workflow and interpret the results? Does it make sense to cross-correlate with 1sliding minute of the file (i.e audio slided with a 30s window)? Thanks $\endgroup$ – JavierG Sep 11 '19 at 20:09
  • $\begingroup$ I'd recommend going through the myriad related questions on this site as a start. $\endgroup$ – MBaz Sep 11 '19 at 20:59

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.