Mel spectogram conversion
#1
by
sahilshah
- opened
It seems to call the model I need to convert the audio into a mel spectogram. In python I have done that with whisper's API however here I am wondering how to go about it? Wondering if there are Apple APIs or reference implementations that are well tested or do I need to write one from scratch?
If you use model from swift, vDSP.FFT on Accelerate can convert fft and convert raw fft to mel by table.
https://developer.apple.com/documentation/accelerate/vdsp/fft