Hugo Farajallah commited on
Commit ·
6cf26ec
1
Parent(s): 90294d8
doc(README): expending the documentation a little.
Browse files
README.md
CHANGED
|
@@ -21,7 +21,15 @@ It is all based on WavLM Base + Phonemizer FR-IT
|
|
| 21 |
|
| 22 |
You need Python 3.12.
|
| 23 |
|
| 24 |
-
Use either uv (recommended), or
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 25 |
|
| 26 |
## main.py
|
| 27 |
|
|
@@ -52,11 +60,3 @@ The advanced settings will let give you fine-grained control over the generated
|
|
| 52 |
It allows to circumvent shortcomings that a simple Arg Max decoder would give you.
|
| 53 |
|
| 54 |

|
| 55 |
-
|
| 56 |
-
## Ideas
|
| 57 |
-
|
| 58 |
-
- [x] Show activation logits of WavLM (fake model for now)
|
| 59 |
-
- [ ] Compare performances with Wav2Vec 2.0-Phonemizer-FR
|
| 60 |
-
- [x] Animate activation logits over time.
|
| 61 |
-
- [x] SHow the result from the feature encoder.
|
| 62 |
-
|
|
|
|
| 21 |
|
| 22 |
You need Python 3.12.
|
| 23 |
|
| 24 |
+
Use either uv (recommended), or pip.
|
| 25 |
+
|
| 26 |
+
```shell
|
| 27 |
+
# Using uv (recommended)
|
| 28 |
+
uv sync
|
| 29 |
+
|
| 30 |
+
# With pip
|
| 31 |
+
pip install -r requirements.txt`.
|
| 32 |
+
```
|
| 33 |
|
| 34 |
## main.py
|
| 35 |
|
|
|
|
| 60 |
It allows to circumvent shortcomings that a simple Arg Max decoder would give you.
|
| 61 |
|
| 62 |

|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|