Generating spectrograms from given base audiofiles

ubclaunchpad / minutes

:telescope: Speaker diarization via transfer learning

27 stars 5 forks source link

Closed grig-guz closed 6 years ago

grig-guz commented 6 years ago

Given a dictionary with speaker ids as keys and lists of file paths as values, generate a spectrogram dataset.

Get data, build the dict, check resulting shapes! Also, try

plt.imshow(X_train[index], aspect = "auto")

to see if spectrograms look ok.

iKevinY commented 6 years ago

@chadlagore implemented a variant of this in 75243456c232e0e5295129b52cf9181f7ec23cd6 as part of the library overhaul in #107.