Ask HN: How would you find “interesting” audio from a video file or stream?
Let's say you have a 12 hour long video file from a surveillance or security camera, and you know that motion detection didn't find any interesting events in the video part. What about the audio? Is there any software that can let you define "interesting" audio, even just amplitude over a certain level. For example, given a minimum threshold in dB and a video file, give me a shortened video file that contains all the video except the periods where the threshold was too low? This is an extremely low-budget solution, but you could load the audio into a program that charts the amplitude of the signal over its length, like most audio editors (e.g. Audacity [1]) Then you can visually see spikes [2]. You may be able to use such an editor to remove the ambient noise and see it more clearly as well. You might be able to hack something up using a audio library such as: