peak_picking with smooth parameter as a NumPy array crashes #247
Description
The docstring states:
smooth : int or numpy array
Smooth the activation function with the kernel (size).
However, the following crashes:
import numpy as np
from madmom.features.onsets import peak_picking
activations = np.random.random_sample((3, 2))
xs, ys = peak_picking(activations, threshold=0.5, smooth=np.array([0, 3]))The error has to do with the NumPy array:
ValueError Traceback (most recent call last)
<ipython-input-66-9fad6d78cba8> in <module>()
1 activations = np.random.random_sample((3, 2))
----> 2 xs, ys = peak_picking(activations, threshold=0.5, smooth=np.array([0, 3]))
/home/carl/anaconda3/lib/python3.6/site-packages/madmom/features/onsets.py in peak_picking(activations, threshold, smooth, pre_avg, post_avg, pre_max, post_max)
919 """
920 # smooth activations
--> 921 if smooth not in (None, 0):
922 activations = smooth_signal(activations, smooth)
923 # compute a moving average
ValueError: The truth value of an array with more than one element is ambiguous. Use a.any() or a.all()
Am I misunderstanding the documentation or is this a bug? Or is it only allowed to pass a 1D array with a single element? Then maybe that should be removed from the docstring entirely, as it's a bit confusing IMO.