Update README.md
Browse files
README.md
CHANGED
|
@@ -57,6 +57,8 @@ python examples/textless_nlp/gslm/speech2unit/clustering/quantize_with_kmeans.py
|
|
| 57 |
unit2speech model is modified Tacotron2 model that learns to synthesize speech from discrete speech units.
|
| 58 |
You can convert the discrete unit to synthesized voice through this [model](https://huggingface.co/nonmetal/gslm-japanese/resolve/main/checkpoint_125k.pt). Also, it is required to download [Waveglow checkpoint](https://dl.fbaipublicfiles.com/textless_nlp/gslm/waveglow_256channels_new.pt) for Vocoder.
|
| 59 |
|
|
|
|
|
|
|
| 60 |
```
|
| 61 |
TTS_MODEL_PATH=<unit2speech_model_file_path>
|
| 62 |
OUT_DIR=<dir_to_dump_synthesized_audio_files>
|
|
|
|
| 57 |
unit2speech model is modified Tacotron2 model that learns to synthesize speech from discrete speech units.
|
| 58 |
You can convert the discrete unit to synthesized voice through this [model](https://huggingface.co/nonmetal/gslm-japanese/resolve/main/checkpoint_125k.pt). Also, it is required to download [Waveglow checkpoint](https://dl.fbaipublicfiles.com/textless_nlp/gslm/waveglow_256channels_new.pt) for Vocoder.
|
| 59 |
|
| 60 |
+
Conversion from unit to speech is available with ```unit2speech_ja.py``` from this repository. It is also required to use ```hparam.py``` for extended compatability.
|
| 61 |
+
|
| 62 |
```
|
| 63 |
TTS_MODEL_PATH=<unit2speech_model_file_path>
|
| 64 |
OUT_DIR=<dir_to_dump_synthesized_audio_files>
|