I am trying to understand the test_lpcnet.py, but there are lots of hardcoded constants for example nb_features = 36 and features[:,:,18:19] ... Why ? What means such numbers and why they are hardcoded ?
In my LPCnet I have nb_features = 42 and also I do not have loops around encoder and decoder ...
What means this formula periods = (.1 + 50*features[:,:,18:19]+100).astype('int16') ??
What mean pcm = np.zeros((nb_frames*pcm_chunk_size, )) ??
Is it possible to add more comments what means some of the constants or to add at least meaningful names ?
Hi all,
I am trying to understand the test_lpcnet.py, but there are lots of hardcoded constants for example
nb_features = 36
andfeatures[:,:,18:19]
... Why ? What means such numbers and why they are hardcoded ?In my LPCnet I have
nb_features = 42
and also I do not have loops around encoder and decoder ... What means this formulaperiods = (.1 + 50*features[:,:,18:19]+100).astype('int16')
?? What meanpcm = np.zeros((nb_frames*pcm_chunk_size, ))
??Is it possible to add more comments what means some of the constants or to add at least meaningful names ?