I've been trying to modify the sound intensity (volume) of a snare sample via python code.
Basically, I used two approaches:
Multiplying the time signal by a negative exponential term, i.e. e^-3t.
Take the Fourier transform, multiplying it by a decimal value lesser than one and applying the inverse Fourier transformation.
Both the results are a noisy unpleasant sound, different of the original one. I plot some graphics to evaluate the wave format, taking a totally inaccurate eye analysis, the data seem similar to the original one.
The code can be found on my github.
I have two questions:
- How can I approach this problem in a better way?
- Would correlation be a good parameter to analyze the similarities between the results and the original sample?