Upload README.md with huggingface_hub
Browse files
README.md
CHANGED
|
@@ -20,8 +20,6 @@ datasets:
|
|
| 20 |
|
| 21 |
This repository hosts the **DiCoW\_v3.2** model developed by [BUT Speech@FIT](https://github.com/BUTSpeechFIT), tailored for **multi-talker automatic speech recognition (MT-ASR)**.
|
| 22 |
|
| 23 |
-
---
|
| 24 |
-
|
| 25 |
## 🔧 Key Improvements over DiCoW v1
|
| 26 |
|
| 27 |
* **FDDT (Frame-Level Diarization Dependent Transformation)** before positional embeddings
|
|
@@ -40,6 +38,7 @@ This repository hosts the **DiCoW\_v3.2** model developed by [BUT Speech@FIT](ht
|
|
| 40 |
|
| 41 |
* Updated **learning schedule**
|
| 42 |
* Improved **hallucination detection & mitigation** during inference
|
|
|
|
| 43 |
---
|
| 44 |
|
| 45 |
|
|
|
|
| 20 |
|
| 21 |
This repository hosts the **DiCoW\_v3.2** model developed by [BUT Speech@FIT](https://github.com/BUTSpeechFIT), tailored for **multi-talker automatic speech recognition (MT-ASR)**.
|
| 22 |
|
|
|
|
|
|
|
| 23 |
## 🔧 Key Improvements over DiCoW v1
|
| 24 |
|
| 25 |
* **FDDT (Frame-Level Diarization Dependent Transformation)** before positional embeddings
|
|
|
|
| 38 |
|
| 39 |
* Updated **learning schedule**
|
| 40 |
* Improved **hallucination detection & mitigation** during inference
|
| 41 |
+
*
|
| 42 |
---
|
| 43 |
|
| 44 |
|