A very similar use of Python & MoviePy to locate the goals in football matches is given and explained, here, with code here.
Given that this was doing for an video file almost exactly what you are asking for with an audio file then I am reasonably sure that you could adapt it to your scenario. Possibly saving as timestamp named clips periods where the average, or peak, volume is above average or a fixed threshold.
After a little trial and error you can almost certainly get a script that will reliably process your audio files for you.
Note that the above is all Free, Open Source & Cross Platform.
