app.py

import gradio as gr
from TTS.api import TTS
import os
from TTS.tts.configs.xtts_config import XttsConfig


# ✅ Accept Coqui License Automatically
os.environ["COQUI_TOS_AGREED"] = "1"
torch.serialization.add_safe_globals([XttsConfig])

# Load the voice cloning model
tts = TTS("tts_models/multilingual/multi-dataset/xtts_v2").to("cpu")

def generate_cloned_voice(text, reference_audio, language):
    output_path = "output.wav"

    # Generate cloned speech with language specification
    tts.tts_to_file(text=text, speaker_wav=reference_audio, file_path=output_path, language=language)

    return output_path

print(tts.languages)

# Create the Gradio interface
interface = gr.Interface(
    fn=generate_cloned_voice,
    inputs=[
        gr.Textbox(label="Enter Translated Text"),
        gr.Audio(label="Upload Reference Audio", type="filepath"),
        gr.Dropdown(["en", "fr", "de", "es", "it"], label="Select Target Language", value="en")
    ],
    outputs=gr.Audio(label="Generated Cloned Voice"),
    title="Free Voice Cloning API",
    description="Upload a sample voice and input text. Select a language, and the system will generate the text in the same voice."
)

# Launch the Gradio app
interface.launch()