app.py

import os
import asyncio
from google import genai
from google.genai import types
import gradio as gr

# Set your Gemini API key (configure via Hugging Face Secrets)
API_KEY = os.getenv("GEMINI_API_KEY")

client = genai.Client(api_key=API_KEY)

async def generate_audio(text):
    config = types.LiveConnectConfig(
        response_modalities=["audio"],
        speech_config=types.SpeechConfig(
            voice_config=types.VoiceConfig(
                prebuilt_voice_config=types.PrebuiltVoiceConfig(voice_name="Puck")
            )
        ),
        system_instruction=types.Content(
            parts=[types.Part.from_text("Repeat user input exactly without explanation")],
            role="user"
        ),
    )
    
    audio_data = b""
    async with client.aio.live.connect(model="models/gemini-2.0-flash-exp", config=config) as session:
        await session.send(input=text, end_of_turn=True)
        async for response in session.receive():
            if data := response.data:
                audio_data += data
    
    # Save as WAV file (16-bit PCM, 24kHz)
    with open("output.wav", "wb") as f:
        f.write(b"RIFF\x00\x00\x00\x00WAVEfmt \x10\x00\x00\x00\x01\x00\x01\x00\x00\x7d\x00\x00\x02\x00\x10\x00data\x00\x00\x00\x00")
        f.write(audio_data)
    
    return "output.wav"

def tts(text):
    if not text.strip():
        return None
    try:
        asyncio.run(generate_audio(text))
        return "output.wav"
    except Exception as e:
        return f"Error: {str(e)}"

# Gradio Interface
iface = gr.Interface(
    fn=tts,
    inputs=gr.Textbox(label="Enter Text", placeholder="Type here..."),
    outputs=gr.Audio(label="TTS Output", type="filepath"),
    examples=["Hello, this is a test.", "How are you today?"],
    title="Gemini TTS Demo",
    description="Convert text to speech using Google's Gemini 2.0 Flash model"
)

if __name__ == "__main__":
    iface.launch()