Just a little cleanup from the last PR, to remove some code specific to my use-case
Note that although I had to replace
wav, sr_ret = librosa.load(vid_path, sr=sr)
assert sr_ret == sr
with
wav, sr_ret = sf.read(vid_path)
To fix a librosa issue of it taking 1 second to load each clip, but it may be a problematic fix as the result is variable sample rate per clip. As such this PR restores the original load_wav as the default, but adds a note to the readme explaining this option.
Just a little cleanup from the last PR, to remove some code specific to my use-case
Note that although I had to replace
with
To fix a librosa issue of it taking 1 second to load each clip, but it may be a problematic fix as the result is variable sample rate per clip. As such this PR restores the original
load_wav
as the default, but adds a note to the readme explaining this option.