python-3.xmachine-learningdeep-learningsignal-processingaudio-processing

Extracting a particular sound from an audio clip


Suppose there is a wav file with sounds such as bird noise, tractor noise, footstep noise all together how to extract the footstep noise alone in python. I want to do this for my project on audio processing to detect footstep noise alone

Applying noise reduction techniques dismisses the footstep audio as a noise. I also don't know what kind of filter is to be used (low/high pass filter) in terms of signal processing


Solution

  • Basically all you need is audio classification.Start with the waveform of audio. If you visualize that waveform you'll get to know that every sound has different pattern. For eg: The graph(or waveform) of dog's bark varies from that of cat's meow. In your case footstep noise, you just need to extract that specific part of graph (for footstep) and treat other sound as noise. You can use "Spectogram" technique, you can apply computer vision technique to the spectogram. You can classify sound with the technique used similar to classify images. Check the spectogram image for your reference