Spaces:
Sleeping
Sleeping
core space code
Browse files
app.py
CHANGED
|
@@ -0,0 +1,90 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import gradio as gr
|
| 2 |
+
from transformers import VitsModel, AutoTokenizer
|
| 3 |
+
import torch
|
| 4 |
+
import logging
|
| 5 |
+
|
| 6 |
+
# Configure logging
|
| 7 |
+
logging.basicConfig(level=logging.INFO)
|
| 8 |
+
logger = logging.getLogger(__name__)
|
| 9 |
+
|
| 10 |
+
# Example text in Bambara
|
| 11 |
+
example = "An filɛ ni ye yɔrɔ minna ni an ye an sigi ka a layɛ yala an bɛ ka baara min kɛ ɛsike a kɛlen don ka Ɲɛ wa ?"
|
| 12 |
+
|
| 13 |
+
# Load model and tokenizer
|
| 14 |
+
try:
|
| 15 |
+
logger.info("Loading Bambara TTS model and tokenizer...")
|
| 16 |
+
model = VitsModel.from_pretrained("sudoping01/bambara-tts")
|
| 17 |
+
tokenizer = AutoTokenizer.from_pretrained("sudoping01/bambara-tts")
|
| 18 |
+
logger.info("Successfully loaded model and tokenizer")
|
| 19 |
+
except Exception as e:
|
| 20 |
+
logger.error(f"Failed to load model: {str(e)}")
|
| 21 |
+
raise Exception(f"Model loading failed: {str(e)}")
|
| 22 |
+
|
| 23 |
+
def generate_audio(text):
|
| 24 |
+
if not text.strip():
|
| 25 |
+
return None, "Please enter some text to synthesize."
|
| 26 |
+
|
| 27 |
+
try:
|
| 28 |
+
inputs = tokenizer(text, return_tensors="pt")
|
| 29 |
+
|
| 30 |
+
with torch.no_grad():
|
| 31 |
+
output = model(**inputs).waveform
|
| 32 |
+
|
| 33 |
+
waveform = output.squeeze().cpu().numpy()
|
| 34 |
+
sample_rate = model.config.sampling_rate
|
| 35 |
+
|
| 36 |
+
return (sample_rate, waveform), None
|
| 37 |
+
except Exception as e:
|
| 38 |
+
logger.error(f"Error during inference: {str(e)}")
|
| 39 |
+
return None, f"Error generating audio: {str(e)}"
|
| 40 |
+
|
| 41 |
+
def load_example():
|
| 42 |
+
return example
|
| 43 |
+
|
| 44 |
+
# Create Gradio interface
|
| 45 |
+
with gr.Blocks(title="Bambara TTS") as demo:
|
| 46 |
+
gr.Markdown(
|
| 47 |
+
"""
|
| 48 |
+
# Bambara TTS: Text-to-Speech for Bambara Language 🇲🇱
|
| 49 |
+
|
| 50 |
+
High-quality text-to-speech for Bambara (Bamanankan), spoken by over 14 million people.
|
| 51 |
+
|
| 52 |
+
- ✅ Real-time TTS with fast response
|
| 53 |
+
- ✅ Runs on CPU
|
| 54 |
+
|
| 55 |
+
## How to Use
|
| 56 |
+
1. Enter your Bambara text or load the example
|
| 57 |
+
2. Click **"Generate Audio"** to listen
|
| 58 |
+
"""
|
| 59 |
+
)
|
| 60 |
+
|
| 61 |
+
with gr.Row():
|
| 62 |
+
with gr.Column():
|
| 63 |
+
text = gr.Textbox(label="Bambara Text", lines=5, placeholder="Type your text in Bambara here...")
|
| 64 |
+
example_btn = gr.Button("Load Example")
|
| 65 |
+
|
| 66 |
+
generate_btn = gr.Button("Generate Audio", variant="primary")
|
| 67 |
+
audio_output = gr.Audio(label="Generated Audio", type="numpy")
|
| 68 |
+
error_msg = gr.Textbox(label="Status", visible=False)
|
| 69 |
+
|
| 70 |
+
# Footer
|
| 71 |
+
gr.Markdown(
|
| 72 |
+
"""
|
| 73 |
+
By [sudoping01](https://huggingface.co/sudoping01). Model: [sudoping01/bambara-tts](https://huggingface.co/sudoping01/bambara-tts). Fine-tuned on Meta's MMS TTS. License: CC BY-NC 4.0 (non-commercial use).
|
| 74 |
+
"""
|
| 75 |
+
)
|
| 76 |
+
|
| 77 |
+
# Connect buttons to functions
|
| 78 |
+
generate_btn.click(
|
| 79 |
+
fn=generate_audio,
|
| 80 |
+
inputs=[text],
|
| 81 |
+
outputs=[audio_output, error_msg]
|
| 82 |
+
)
|
| 83 |
+
example_btn.click(
|
| 84 |
+
fn=load_example,
|
| 85 |
+
inputs=[],
|
| 86 |
+
outputs=text
|
| 87 |
+
)
|
| 88 |
+
|
| 89 |
+
# Launch the demo
|
| 90 |
+
demo.launch()
|