app.py

import gradio as gr
import torchaudio
from tortoise.api import TextToSpeech
from tortoise.utils.audio import load_voice

# Initialize TorToiSe
tts = TextToSpeech()

# List of available voices
AVAILABLE_VOICES = [
    "angie", "daniel", "deniro", "emma", "freeman", "geralt", "halle", 
    "jlaw", "lj", "myself", "pat", "snakes", "tom", "train_atkins", 
    "train_dotrice", "train_kennard", "weaver", "william"
]

# Function to generate speech
def generate_speech(text, voice):
    if not text:
        return "Please enter some text."
    
    try:
        # Load the selected voice
        voice_samples, conditioning_latents = load_voice(voice)
        
        # Generate speech
        gen = tts.tts_with_preset(
            text, 
            voice_samples=voice_samples, 
            conditioning_latents=conditioning_latents, 
            preset="fast"
        )
        
        # Save the output
        output_file = "output.wav"
        torchaudio.save(output_file, gen.squeeze(0).cpu(), 24000)
        
        return output_file
    except Exception as e:
        return f"Error generating speech: {str(e)}"

# Gradio interface
with gr.Blocks(
    theme=gr.themes.Soft(primary_hue="teal", secondary_hue="pink"),
    css=".gradio-container {background: linear-gradient(135deg, #f5f7fa, #c3cfe2);} "
        "button {background: linear-gradient(135deg, #6a11cb, #2575fc); color: white; border: none; padding: 10px 20px; border-radius: 5px;} "
        "button:hover {background: linear-gradient(135deg, #2575fc, #6a11cb);} "
) as demo:
    # Title and description
    gr.Markdown(
        """
        # 🎤 **TorToiSe Text-to-Speech**
        **Generate natural-sounding speech in multiple voices!**
        """
    )

    # Inputs
    with gr.Row():
        text_input = gr.Textbox(
            lines=5,
            label="📝 **Enter Text**",
            placeholder="Type your text here..."
        )
        voice_input = gr.Dropdown(
            choices=AVAILABLE_VOICES,
            label="🗣️ **Select Voice**",
            value="emma"  # Default voice
        )

    # Output
    output_audio = gr.Audio(label="🎧 **Generated Speech**", type="filepath")

    # Submit button
    submit_button = gr.Button("✨ **Generate Speech**")

    # Link button to function
    submit_button.click(
        generate_speech,
        inputs=[text_input, voice_input],
        outputs=output_audio
    )

# Launch the app
demo.launch()