app.py

import os
import uuid
import whisper
from groq import Groq
import gradio as gr
from gtts import gTTS

# -------------------------------
# Groq API Key (set in HF Secrets)
# -------------------------------
client = Groq(api_key=os.environ["GROQ_API_KEY"])

# -------------------------------
# Load Whisper model
# -------------------------------
stt_model = whisper.load_model("base")

# -------------------------------
# Voice-to-Voice function
# -------------------------------
def voice_to_voice(audio_path):
    try:
        # 1️⃣ Speech-to-Text
        text = stt_model.transcribe(audio_path)["text"]

        # 2️⃣ LLM response via Groq
        response = client.chat.completions.create(
            model="llama-3.3-70b-versatile",
            messages=[{"role": "user", "content": text}]
        ).choices[0].message.content

        # 3️⃣ Text-to-Speech (gTTS)
        output_file = f"out_{uuid.uuid4().hex}.mp3"
        tts = gTTS(text=response, lang='en')
        tts.save(output_file)

        return output_file

    except Exception as e:
        return f"Error: {str(e)}"

# -------------------------------
# Gradio UI
# -------------------------------
interface = gr.Interface(
    fn=voice_to_voice,
    inputs=gr.Audio(type="filepath", label="Speak"),
    outputs=gr.Audio(type="filepath", label="Assistant Reply"),
    title="🎙️ Voice-to-Voice AI Assistant",
    description="Whisper → Groq LLaMA-3 → Google TTS (Free & Open Source)"
)

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