Spaces:
Runtime error
Runtime error
Create app.py
Browse files
app.py
ADDED
@@ -0,0 +1,69 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import gradio as gr
|
2 |
+
import torch
|
3 |
+
from transformers import AutoProcessor, AutoModel
|
4 |
+
import scipy.io.wavfile as wavfile
|
5 |
+
import spaces # Import the spaces module
|
6 |
+
|
7 |
+
# Load the model and processor
|
8 |
+
def load_model():
|
9 |
+
processor = AutoProcessor.from_pretrained("suno/bark-small")
|
10 |
+
model = AutoModel.from_pretrained("suno/bark-small")
|
11 |
+
model.eval() # Set the model to evaluation mode
|
12 |
+
return processor, model
|
13 |
+
|
14 |
+
# Load models on startup
|
15 |
+
print("Loading models...")
|
16 |
+
processor, model = load_model()
|
17 |
+
print("Models loaded successfully!")
|
18 |
+
|
19 |
+
@spaces.GPU # Decorate the function to enable GPU usage
|
20 |
+
def text_to_speech(text):
|
21 |
+
try:
|
22 |
+
# Check if a GPU is available and set device
|
23 |
+
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
|
24 |
+
|
25 |
+
# Move model to GPU
|
26 |
+
model.to(device)
|
27 |
+
|
28 |
+
inputs = processor(
|
29 |
+
text=[text],
|
30 |
+
return_tensors="pt",
|
31 |
+
).to(device) # Move inputs to GPU
|
32 |
+
|
33 |
+
# Generate speech values on the GPU
|
34 |
+
with torch.no_grad(): # Disable gradient calculation for inference
|
35 |
+
speech_values = model.generate(**inputs, do_sample=True)
|
36 |
+
|
37 |
+
# Move generated audio data back to CPU for saving
|
38 |
+
audio_data = speech_values.cpu().numpy().squeeze()
|
39 |
+
sampling_rate = model.generation_config.sample_rate
|
40 |
+
|
41 |
+
temp_path = "temp_audio.wav"
|
42 |
+
wavfile.write(temp_path, sampling_rate, audio_data)
|
43 |
+
|
44 |
+
return temp_path
|
45 |
+
except Exception as e:
|
46 |
+
return f"Error generating speech: {str(e)}"
|
47 |
+
|
48 |
+
# Define Gradio interface
|
49 |
+
demo = gr.Interface(
|
50 |
+
fn=text_to_speech,
|
51 |
+
inputs=[
|
52 |
+
gr.Textbox(
|
53 |
+
label="Enter text (Hindi supported)",
|
54 |
+
placeholder="दिल्ली मेट्रो में आपका स्वागत है"
|
55 |
+
)
|
56 |
+
],
|
57 |
+
outputs=gr.Audio(label="Generated Speech"),
|
58 |
+
title="Bark TTS Test App",
|
59 |
+
description="This app generates speech from text using the Bark TTS model. Supports Hindi.",
|
60 |
+
examples=[
|
61 |
+
["दिल्ली मेट्रो में आपका स्वागत है"],
|
62 |
+
["कृपया ध्यान दें"],
|
63 |
+
["अगला स्टेशन राजीव चौक है"]
|
64 |
+
],
|
65 |
+
theme="compact" # You can customize the theme
|
66 |
+
)
|
67 |
+
|
68 |
+
if __name__ == "__main__":
|
69 |
+
demo.launch()
|