search

andabi / music-source-separation

Deep neural networks for separating singing voice from music written in TensorFlow
795 stars 150 forks source link

to_wav_mag_only bug? #35

Open sniperwrb opened 5 years ago

sniperwrb commented 5 years ago

python 3.6 tensorflow 1.13

in preprocess.py, function "to_wav_mag_only": return np.array(list(map(lambda m: lambda p: griffin_lim(m, len_frame, len_hop, num_iters=num_iters, phase_angle=p), list(zip(mag, init_phase))[1]))) is buggy

change to: return np.array(list(map(lambda m,p: griffin_lim(m, len_frame, len_hop, num_iters=num_iters, phase_angle=p), mag, init_phase))) works, but i don't know if it affects the result..