search

pyannote / pyannote-audio

Neural building blocks for speaker diarization: speech activity detection, speaker change detection, overlapped speech detection, speaker embedding
http://pyannote.github.io
MIT License
5.5k stars 726 forks source link

"Unable to allocate 81.3 GiB" #1257

Closed doublex closed 10 months ago

doublex commented 1 year ago

This ogg-file: 1525832.zip

Testcase:

import pyannote.audio.pipelines
pipeline = pyannote.audio.pipelines.SpeakerSegmentation( segmentation="pyannote/segmentation@Interspeech2021", use_auth_token="hf_123456789" )
pipeline.instantiate({
    'onset': 0.84,
    'offset': 0.46,
    'min_duration_on': 0.0,
    'min_duration_off': 0.0,
    'stitch_threshold': 0.39,
})
speaker_segmentation = pipeline( '1525832.ogg' )
for turn, _, speaker in speaker_segmentation.itertracks(yield_label=True):
    print( turn.start, turn.end, speaker )

Console:

Traceback (most recent call last):
  File "/mnt/cuda/testpyannote.py", line 11, in <module>
    speaker_segmentation = pipeline( '1525832.wav' )
  File "/home/cuda/.local/lib/python3.9/site-packages/pyannote/audio/core/pipeline.py", line 238, in __call__
    return self.apply(file, **kwargs)
  File "/home/cuda/.local/lib/python3.9/site-packages/pyannote/audio/pipelines/segmentation.py", line 307, in apply
    stitched_segmentations = np.NAN * np.zeros(
numpy.core._exceptions._ArrayMemoryError: Unable to allocate 81.3 GiB for an array with shape (8776, 263, 4726) and data type float64
stale[bot] commented 11 months ago

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.