Instructions to use Kyudan/eDIAR with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use Kyudan/eDIAR with Transformers:
# pip install -U transformers accelerate # Load model directly from transformers import AutoModel model = AutoModel.from_pretrained("Kyudan/eDIAR", device_map="auto") - Notebooks
- Google Colab
- Kaggle
Remove code link and leader-follower row
Browse files
README.md
CHANGED
|
@@ -23,7 +23,7 @@ official [e-means](https://github.com/facebookresearch/faiss/tree/main/contrib/e
|
|
| 23 |
The streaming system re-clusters a bounded 90 s window every chunk, may revise labels for 10 s, and anchors permanent
|
| 24 |
speaker ids to the slots of the head.
|
| 25 |
|
| 26 |
-
Demo: https://kyudan1.github.io/eDIAR/
|
| 27 |
|
| 28 |
## Usage: streaming or non-streaming
|
| 29 |
|
|
@@ -78,7 +78,6 @@ Install with `pip install -r requirements.txt`.
|
|
| 78 |
|---|---|---|---|---|
|
| 79 |
| Streaming Sortformer 4spk v2.1 | 14.1 | 8.3 | 27.7 | 41.4 |
|
| 80 |
| Nemotron-3 head (8 slots) | 6.4 | 9.3 | 27.4 | 19.8 |
|
| 81 |
-
| Online leader-follower | 5.9 | 10.7 | 31.6 | 11.6 |
|
| 82 |
| **This repository** (after revision) | 5.8 | 9.1 | 27.4 | 16.4 |
|
| 83 |
|
| 84 |
Numbers are generated from the result files of the accompanying paper (cropped sessions of 7-8 minutes; 44 / 16 / 15 / 18 sessions).
|
|
|
|
| 23 |
The streaming system re-clusters a bounded 90 s window every chunk, may revise labels for 10 s, and anchors permanent
|
| 24 |
speaker ids to the slots of the head.
|
| 25 |
|
| 26 |
+
Demo: https://kyudan1.github.io/eDIAR/
|
| 27 |
|
| 28 |
## Usage: streaming or non-streaming
|
| 29 |
|
|
|
|
| 78 |
|---|---|---|---|---|
|
| 79 |
| Streaming Sortformer 4spk v2.1 | 14.1 | 8.3 | 27.7 | 41.4 |
|
| 80 |
| Nemotron-3 head (8 slots) | 6.4 | 9.3 | 27.4 | 19.8 |
|
|
|
|
| 81 |
| **This repository** (after revision) | 5.8 | 9.1 | 27.4 | 16.4 |
|
| 82 |
|
| 83 |
Numbers are generated from the result files of the accompanying paper (cropped sessions of 7-8 minutes; 44 / 16 / 15 / 18 sessions).
|