metadata
license: mit
base_model: pyannote/segmentation-3.0
tags:
- speaker-diarization
- speaker-segmentation
- generated_from_trainer
datasets:
- diarizers-community/voxconverse
model-index:
- name: speaker-segmentation-fine-tuned-voxconverse-en
results: []
speaker-segmentation-fine-tuned-voxconverse-en
This model is a fine-tuned version of pyannote/segmentation-3.0 on the diarizers-community/voxconverse dataset. It achieves the following results on the evaluation set:
- Loss: 1.1250
- Der: 0.8257
- False Alarm: 0.3733
- Missed Detection: 0.3995
- Confusion: 0.0528
Model description
More information needed
Intended uses & limitations
More information needed
Training and evaluation data
More information needed
Training procedure
Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 0.001
- train_batch_size: 32
- eval_batch_size: 32
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: cosine
- num_epochs: 5.0
Training results
| Training Loss | Epoch | Step | Validation Loss | Der | False Alarm | Missed Detection | Confusion |
|---|---|---|---|---|---|---|---|
| 0.9302 | 1.0 | 791 | 0.9903 | 0.6790 | 0.5013 | 0.0965 | 0.0812 |
| 0.8848 | 2.0 | 1582 | 1.0536 | 0.7965 | 0.3991 | 0.3409 | 0.0565 |
| 0.8513 | 3.0 | 2373 | 1.0884 | 0.8114 | 0.4017 | 0.3528 | 0.0569 |
| 0.7926 | 4.0 | 3164 | 1.1292 | 0.8378 | 0.3660 | 0.4219 | 0.0500 |
| 0.8147 | 5.0 | 3955 | 1.1250 | 0.8257 | 0.3733 | 0.3995 | 0.0528 |
Framework versions
- Transformers 4.40.1
- Pytorch 2.2.0+cu121
- Datasets 2.17.0
- Tokenizers 0.19.1