search

ina-foss / inaSpeechSegmenter

CNN-based audio segmentation toolkit. Allows to detect speech, music, noise and speaker gender. Has been designed for large scale gender equality studies based on speech time per gender.
MIT License
736 stars 126 forks source link

memory leak #42

Closed tomjerrygithub closed 1 year ago

tomjerrygithub commented 4 years ago

Line # Mem usage Increment Line Contents

 6    363.0 MiB    363.0 MiB   @profile
 7                             def test():
 8                                 
 9    363.0 MiB      0.0 MiB       media = '8560_00.mp4'
10                             
11    710.4 MiB    347.4 MiB       segmentation = seg(media)
12                                 
13    710.4 MiB      0.0 MiB       print(segmentation)

Line # Mem usage Increment Line Contents

 6    937.0 MiB    937.0 MiB   @profile
 7                             def test():
 8                                 
 9    937.0 MiB      0.0 MiB       media = '8560_00.mp4'
10                             
11   1086.5 MiB    149.6 MiB       segmentation = seg(media)
12                                 
13   1086.5 MiB      0.0 MiB       print(segmentation)

Line # Mem usage Increment Line Contents

 6   1498.8 MiB   1498.8 MiB   @profile
 7                             def test():
 8                                 
 9   1498.8 MiB      0.0 MiB       media = '8560_00.mp4'
10                             
11   1613.3 MiB    114.5 MiB       segmentation = seg(media)
12                                 
13   1613.3 MiB      0.0 MiB       print(segmentation)

` from inaSpeechSegmenter import Segmenter, seg2csv from memory_profiler import profile

seg = Segmenter(vad_engine='smn',detect_gender=False)

@profile def test():

media = '8560_00.mp4'

segmentation = seg(media)

print(segmentation)

for i in range(10): test() `

python 3.7.3 tensorflow 2.2.0 keras 2.4.3

DavidDoukhan commented 1 year ago

Dear @tomjerrygithub ,

These issues were due to changes in tensorflow behavior. These have been fixed.

chazo1994 commented 10 months ago

changes in tensorflow

@DavidDoukhan Which version of tensorflow or the commit of this repos was fixed these issues?