0

I have used FFMPEG to extract decibel (or rms? I am not familiar with the units) values of the audio volume from an mp4. I have 20 samples per frame.

enter image description here

How can I use these values (which are negative in almost all frames), to determine if the frame is silent or has audio (music, speech, etc)?

Abbas
  • 1,118
  • 1
  • 11
  • 25
  • This is a question better pointed at dsp.stackexchange.com , but to give a little flavour: Decibel is the unit and in this case [it is dBFS](https://dsp.stackexchange.com/a/8786/49408). RMS is the measure and is a way of expressing loudness. By nature dBFS starts at `0` and goes down from there. Absolute digital silence should be -∞, NaN. "Really really quiet" will depend on the format but RMS = -80.0 dBFS would be a good benchmark for 44.1kHz / 16-bit PCM. – fdcpp Jun 26 '21 at 21:17
  • 1
    The silencedetect filter uses a threshold of -60 dB. – Gyan Jun 27 '21 at 04:13
  • https://ffmpeg.org/ffmpeg-filters.html#silencedetect – fdcpp Jun 27 '21 at 08:30

0 Answers0