onsetsAubio

maelzel.snd.features.onsetsAubio(samples, sr, method='mkl', winsize=1024, hopsize=512, threshold=0.03, mingap=0.05, silencedb=-70)[source]

Detect onsets in samples

Parameters:

  • samples (ndarray) – the samples, as numpy array (1D), between -1 and 1

  • sr (int) – the sample rate of samples

  • winsize – the size of the fft window size, in samples

  • hopsize – the hop size, in samples

  • threshold – depends on the method. The lower this value, the more probable is it that an onset is detected

  • method – the method to detect onsets. One of: - energy: local energy, - hfc: high frequency content, - complex: complex domain, - phase: phase-based method, - wphase: weighted phase deviation, - specdiff: spectral difference, - kl: Kullback-Liebler, - mkl: modified Kullback-Liebler, - specflux: spectral flux.

  • mingap – the min. amount of time (in seconds) between two onsets

  • silencedb – onsets will only be detected if the amplitude exceeds this value (in dB)

Return type:

list[float]

Returns:

a list of floats, representing the times of the onsets