Transformers
TensorBoard
Safetensors
Hindi
pyannet
speaker-diarization
speaker-segmentation
Generated from Trainer
Instructions to use shreyaskal3/speaker-segmentation-fine-tuned-hindi with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use shreyaskal3/speaker-segmentation-fine-tuned-hindi with Transformers:
# Load model directly from transformers import AutoModel model = AutoModel.from_pretrained("shreyaskal3/speaker-segmentation-fine-tuned-hindi", device_map="auto") - Notebooks
- Google Colab
- Kaggle
kale commited on
Rename config.json to config.yaml
Browse files- config.json +0 -18
- config.yaml +19 -0
config.json
DELETED
|
@@ -1,18 +0,0 @@
|
|
| 1 |
-
{
|
| 2 |
-
"architectures": [
|
| 3 |
-
"SegmentationModel"
|
| 4 |
-
],
|
| 5 |
-
"chunk_duration": 10.0,
|
| 6 |
-
"max_speakers_per_chunk": 3,
|
| 7 |
-
"max_speakers_per_frame": 2,
|
| 8 |
-
"min_duration": null,
|
| 9 |
-
"model_type": "pyannet",
|
| 10 |
-
"sample_rate": 16000,
|
| 11 |
-
"torch_dtype": "float32",
|
| 12 |
-
"transformers_version": "4.47.1",
|
| 13 |
-
"warm_up": [
|
| 14 |
-
0.0,
|
| 15 |
-
0.0
|
| 16 |
-
],
|
| 17 |
-
"weigh_by_cardinality": false
|
| 18 |
-
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
config.yaml
ADDED
|
@@ -0,0 +1,19 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version: 3.1.0
|
| 2 |
+
|
| 3 |
+
pipeline:
|
| 4 |
+
name: pyannote.audio.pipelines.SpeakerDiarization
|
| 5 |
+
params:
|
| 6 |
+
clustering: AgglomerativeClustering
|
| 7 |
+
embedding: pyannote/wespeaker-voxceleb-resnet34-LM
|
| 8 |
+
embedding_batch_size: 32
|
| 9 |
+
embedding_exclude_overlap: true
|
| 10 |
+
segmentation: pyannote/segmentation-3.0
|
| 11 |
+
segmentation_batch_size: 32
|
| 12 |
+
|
| 13 |
+
params:
|
| 14 |
+
clustering:
|
| 15 |
+
method: centroid
|
| 16 |
+
min_cluster_size: 12
|
| 17 |
+
threshold: 0.7045654963945799
|
| 18 |
+
segmentation:
|
| 19 |
+
min_duration_off: 0.0
|