signal processing for audio file in python

Question

I am tying to do silence detection in uncompressed AIFF audio files. I prefer to do it in Python, but would consider other options if this is super inefficient. The uncompressed files I am dealing with are expected to be 20 MB (maximum size).

I can understand basics of signal processing, but am not an expert in it.

Answer 1

You're in luck! The aifc library seems to do enough to support the solving of your problem.

Answer 2

Language-agnostic pseudo code:

• for each time window (e.g. 10 ms)
  • calculate RMS power in time window
  • silence = RMS power < silence threshold

To calculate RMS power:

• sum_sq = 0
• for each sample in N sample window
  • sum_sq += sample^2
• RMS power = sqrt(sum_sq / N)

You probably also want to add a further layer of detection, e.g. decide that silence = M consecutive silent windows, where M determines how long a silence needs to be before it counts as an actual silence.