Auto split of a file according a sound pattern

I recorded several large audio wav files. I need to split them automaticaly into several smaller files. Each original file contains a regular sound pattern (something like a gong in the begining of each chapter). Is there any way how to detect this pattern (similarly like “noise reduction”) in all file and cut it or put a marker on its position? In short, I need one or the second way:

a)to choose a sound pattern and analyze all file and split it everytime the patterns occurs and save it in MP3 format


b) to choose a sound pattern (gong) and analyze all file and on every occurence of the pattern place a marker.


c) if audacity is not capable to do it, could you recommend other software which can do it?

Silence detection function doesn’t work well - the record is spoken speech, so there are regular silences due to the intonation of the speech.

Any idea?

Thank in advance.

You may need different software. Audacity can’t recognize sound patterns and make decisions. Sound events are usually a lot more complicated than anyone thinks and you need a human somewhere in the process to do the recognition.

There was a recent post where someone wanted to be able to recognize the difference between a male and female voice. Even that was harder than you think because most of the energy in human voices is common. Again snapping us back to a human who can tell immediately, but not so the machines.


I didn’t want some AI magics. I just want:

  1. select pattern
  2. Copy to clipboard (or into something like that)
  3. Tell the Audacity "Find this pattern in all wav) - e.g. similar like features “Noise elimination” there also you will take a pattern and let analyze all record and remove all noise accordint that pattern.

Is it clearer?

The main problem is that the actual audio samples are not identical. You can record the exact same thing twice and the digital data won’t be identical because the samples are taken “randomly” on the analog time line. It’s not really “random” but it’s not correlated to the analog. (See [u]Digital Audio Fundamentals[/u].)

And in the real world the analog is never identical… For example, you can digitally-record yourself saying “hello” and make an exact digital copy of that, and if you subtract you’ll get silence. But if you record yourself saying “hello” twice there is a LOT of difference at the digital-sample level. Subtraction sounds exactly like addition (it sounds like a regular “mix” of both recordings… like a pair of twins saying “hello” together).