Update README.md
Browse files
README.md
CHANGED
|
@@ -7,3 +7,25 @@ base_model: Jackellie/ellie-Bert-VITS2
|
|
| 7 |
---
|
| 8 |
|
| 9 |
Taiwan accent TTS model from JackEllie.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 7 |
---
|
| 8 |
|
| 9 |
Taiwan accent TTS model from JackEllie.
|
| 10 |
+
|
| 11 |
+
## Usage
|
| 12 |
+
|
| 13 |
+
Using this checkpoint from Hugging Face Transformers:
|
| 14 |
+
|
| 15 |
+
```python
|
| 16 |
+
from transformers import AutoModel, AutoProcessor
|
| 17 |
+
from scipy.io.wavfile import write
|
| 18 |
+
import torch
|
| 19 |
+
|
| 20 |
+
model_name = "BricksDisplay/ellie-Bert-VITS2"
|
| 21 |
+
|
| 22 |
+
model = AutoModel.from_pretrained(model_name, trust_remote=True)
|
| 23 |
+
processor = AutoProcessor.from_pretrained(model_name, trust_remote=True)
|
| 24 |
+
|
| 25 |
+
with torch.no_grad():
|
| 26 |
+
inputs = processor("你好", language="zh", return_tensors="pt")
|
| 27 |
+
result = model(**inputs)
|
| 28 |
+
result = result["waveform"]
|
| 29 |
+
write("output.wav", model.config.sampling_rate, result[0].numpy())
|
| 30 |
+
|
| 31 |
+
```
|