|
--- |
|
tags: |
|
- pyannote |
|
- pyannote-audio |
|
- pyannote-audio-pipeline |
|
--- |
|
|
|
```python |
|
|
|
# load pretrained pipeline |
|
from pyannote.audio import Pipeline |
|
pipeline = Pipeline.from_pretrained('hbredin/utter-project-diarization') |
|
|
|
# send it to MPS device (on Apple Silicon) |
|
import torch |
|
mps = torch.device('mps') |
|
pipeline.to(mps) |
|
|
|
# apply it on sample file |
|
from pyannote.audio.sample import SAMPLE_FILE |
|
diarization = pipeline(SAMPLE_FILE) |
|
|
|
# print output |
|
print(diarization) |
|
# [ 00:00:06.730 --> 00:00:07.185] A 1 |
|
# [ 00:00:07.590 --> 00:00:08.316] B 1 |
|
# [ 00:00:08.316 --> 00:00:09.852] C speaker90 |
|
# [ 00:00:09.852 --> 00:00:09.902] D 1 |
|
# [ 00:00:09.902 --> 00:00:10.982] E speaker91 |
|
# [ 00:00:10.459 --> 00:00:10.527] F 1 |
|
# [ 00:00:10.527 --> 00:00:14.729] G speaker90 |
|
# [ 00:00:14.307 --> 00:00:17.884] H speaker91 |
|
# [ 00:00:18.019 --> 00:00:21.495] I 3 |
|
# [ 00:00:18.239 --> 00:00:18.374] J speaker91 |
|
# [ 00:00:21.765 --> 00:00:28.515] K speaker91 |
|
# [ 00:00:27.824 --> 00:00:29.984] L speaker90 |
|
|
|
# compute diarization error rate |
|
from pyannote.metrics.diarization import DiarizationErrorRate |
|
metric = DiarizationErrorRate() |
|
metric(SAMPLE_FILE['annotation'], diarization, detailed=True) |
|
# {'missed detection': 0.6146562499999995, |
|
# 'correct': 19.108875, |
|
# 'false alarm': 0.48028125, |
|
# 'confusion': 4.626468749999998, |
|
# 'total': 24.349999999999998, |
|
# 'diarization error rate': 0.23496534907597527} |
|
``` |
|
|
