Update app.py
Browse files
app.py
CHANGED
|
@@ -130,11 +130,16 @@ with gr.Blocks(title="Orpheus Text-to-Speech") as demo:
|
|
| 130 |
gr.Markdown("""
|
| 131 |
# 🎵 [Tajik Orpheus Text-to-Speech](https://huggingface.co/re-skill/orpheus-tj-early)
|
| 132 |
|
| 133 |
-
Enter your text below and hear it converted to natural-sounding speech with the Orpheus TTS model.
|
| 134 |
|
| 135 |
## Tips for better prompts:
|
| 136 |
- Short text prompts generally work better than very long phrases
|
| 137 |
- Increasing `repetition_penalty` and `temperature` makes the model speak faster.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 138 |
""")
|
| 139 |
with gr.Row():
|
| 140 |
with gr.Column(scale=3):
|
|
|
|
| 130 |
gr.Markdown("""
|
| 131 |
# 🎵 [Tajik Orpheus Text-to-Speech](https://huggingface.co/re-skill/orpheus-tj-early)
|
| 132 |
|
| 133 |
+
Enter your text below and hear it converted to natural-sounding speech with the Orpheus TTS model.
|
| 134 |
|
| 135 |
## Tips for better prompts:
|
| 136 |
- Short text prompts generally work better than very long phrases
|
| 137 |
- Increasing `repetition_penalty` and `temperature` makes the model speak faster.
|
| 138 |
+
|
| 139 |
+
## Note:
|
| 140 |
+
- This is demo of early checkpoint trained only on `35 Hours` of data.
|
| 141 |
+
- The model was not fine-tuned on a specific voice. Hence, you will get different voices every time you run the model.
|
| 142 |
+
|
| 143 |
""")
|
| 144 |
with gr.Row():
|
| 145 |
with gr.Column(scale=3):
|