Update README.md
Browse files
README.md
CHANGED
|
@@ -26,8 +26,11 @@ This model is adapted from **[Wav2vec-Classroom](https://huggingface.co/aadel4/W
|
|
| 26 |
|
| 27 |
This model was originally trained using the fairseq library then ported into Huggingface.
|
| 28 |
|
|
|
|
|
|
|
| 29 |
**Use Case:**
|
| 30 |
- Speech-to-text transcription for classroom environments.
|
|
|
|
| 31 |
- Educational research and analysis of classroom discourse.
|
| 32 |
- Low-resource ASR applications where gold-standard labels are limited.
|
| 33 |
|
|
|
|
| 26 |
|
| 27 |
This model was originally trained using the fairseq library then ported into Huggingface.
|
| 28 |
|
| 29 |
+
The model should be run with n-gram LM beamsearch decoding for best results. We got our best results usig [this](https://drive.google.com/drive/u/0/folders/1yAFXcbozqDUFZu-hnnzFP_8SAzDYT2JJ) 5-gram LM we trained on classroom speech text.
|
| 30 |
+
|
| 31 |
**Use Case:**
|
| 32 |
- Speech-to-text transcription for classroom environments.
|
| 33 |
+
- Forced allignment of transcription with audio to provide character and word level boundaries.
|
| 34 |
- Educational research and analysis of classroom discourse.
|
| 35 |
- Low-resource ASR applications where gold-standard labels are limited.
|
| 36 |
|