Skip to content

Error while fine-tuning hyperparameters in the Pyannote.audio 2.1 Speaker Diarization Pipeline #47

@nalli-hu

Description

@nalli-hu

I followed the notebook "Adapting pyannote.audio 2.1 pretrained speaker diarization pipeline to your own data" to adapt the Pyannote.audio 2.1 Speaker Diarization Pipeline to my own data.

I am using the pyannote.database structure to load my data, and the audio recordings are short snippets, with the shortest ones being one second in length.

However, when I try to fine-tune the hyperparameters using the following code:
iterations = optimizer.tune_iter(dev_set, show_progress=False)
best_loss = 1.0
for i, iteration in enumerate(iterations):
print(f"Best segmentation threshold so far: {iteration['params']['segmentation']['threshold']}")
if i > 20: break # 50 iterations should give slightly better results

I encounter the following error:
Traceback (most recent call last): File "/opt/miniconda3/envs/Asr2/lib/python3.9/site-packages/optuna/study/_optimize.py", line 200, in _run_trial value_or_values = func(trial) File "path/to/optimizer.py", line 210, in objective output = pipeline(input) File "/opt/miniconda3/envs/Asr2/lib/python3.9/site-packages/pyannote/audio/core/pipeline.py", line 238, in call return self.apply(file, **kwargs) File "/opt/miniconda3/envs/Asr2/lib/python3.9/site-packages/pyannote/audio/pipelines/speaker_diarization.py", line 494, in apply hardclusters, = self.clustering( File "/opt/miniconda3/envs/Asr2/lib/python3.9/site-packages/pyannote/audio/pipelines/clustering.py", line 612, in call oracle_segmentations = oracle_segmentation(file, window, frames=frames) File "/opt/miniconda3/envs/Asr2/lib/python3.9/site-packages/pyannote/audio/pipelines/utils/oracle.py", line 105, in oracle_segmentation return SlidingWindowFeature(np.float32(np.stack(segmentations)), window) File "<array_function internals>", line 180, in stack File "/opt/miniconda3/envs/Asr2/lib/python3.9/site-packages/numpy/core/shape_base.py", line 422, in stack raise ValueError('need at least one array to stack') ValueError: need at least one array to stack

Any suggestions on how to fix this error would be greatly appreciated.

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions