Closed orasanen closed 4 years ago
on run_SylNet.py line 98 to 100 it reads:
fs, y = scipy.io.wavfile.read(fileList[i])
y = y/max(abs(y))
y = librosa.core.resample(y=y, orig_sr=fs, target_sr=Fs)
my suggestion:
y, _ = librosa.core.load(fileList[i], sr = Fs, mono = True)
y = y/max(abs(y))
This way you are resampling and assuring mono in one line
Hi! Thanks for a great suggestion! This was an issue with SylNet submodule. I have now updated SylNet and added the new version as a submodule to ALICE.
Please note that ALICE moved from "master" to "new_diarizer" branch, and the update only applies to the new branch.
super !
should I git checkout new_diarizer
and git submodules update
?
-- leandrogarber.info
El mar., 26 may. 2020 a las 12:15, orasanen (notifications@github.com) escribió:
Closed #2 https://github.com/orasanen/ALICE/issues/2.
— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/orasanen/ALICE/issues/2#event-3374043077, or unsubscribe https://github.com/notifications/unsubscribe-auth/AAWH6ONQ63NYQKWOUBVA5HDRTPMJJANCNFSM4MZSMYHQ .
I'm not actually sure if that would work. A lot of things were changed in the main repo + both submodules as I upgraded to an entirely new version of the diarizer at the same time. Also, the whole pyannote tools have to be reinstalled from scratch (see the new ALICE readme).
I recommend just reinstalling the entire thing from scratch (including the Conda environment), if you want to change the branch.
Note that the counts from the demo script have also changed, so you can use that to check if things are working correctly or not.
This needs to be fixed.