Use audio analysis to trim movie.

Hi there,

I’m doing a project which involves looking at large numbers of video clips with audio where people speak a single word.

I would like to find the points where they start and stop speaking, and record them in a table to be used to trim the clips to those sections during playback.

So far I have been using logic chops to look at when voice volume goes over a certain threshold and using some lag, so multi-syllable words aren’t split up. This gives me a great realtime analysis of the audio. But I need to reduce this to two numbers that I can record in a table next to the filename of the clip. The frame index number where the volume goes over the threshold, and another where it returns below the threshold.
Is there a method for doing this?

Many thanks

made this thoroughly inelegant patch as proof of concept.

hit the button and speak. it records for 3 secs. first and last frames over a threshold are marked.

needs some work.
audio_trimmer2.5.toe (12.8 KB)