app.py

import os
from pathlib import Path
import gradio as gr
from groq import Groq

# Load API key from Hugging Face Secret
api_key = os.getenv("GroqApiKey")

# Initialize Groq client
client = Groq(api_key=api_key)


def ask_ai(audio_file):
    try:
        # 1. Speech-to-Text
        with open(audio_file, "rb") as file:
            transcription = client.audio.transcriptions.create(
                file=("user_input.wav", file.read()),
                model="whisper-large-v3",
                response_format="verbose_json",
            )
        user_text = transcription.text

        # 2. LLM Completion
        completion = client.chat.completions.create(
            model="llama-3.1-8b-instant",
            messages=[{"role": "user", "content": user_text}],
            temperature=1,
            max_completion_tokens=512,
            top_p=1,
        )
        answer_text = completion.choices[0].message.content

        # 3. Text-to-Speech
        speech_file_path = Path("answer.wav")
        response = client.audio.speech.create(
            model="playai-tts",
            voice="Calum-PlayAI",
            response_format="wav",
            input=answer_text,
        )

        # Save audio file
        with open(speech_file_path, "wb") as f:
            for chunk in response.iter_bytes():
                f.write(chunk)

        return user_text, answer_text, str(speech_file_path)

    except Exception as e:
        return "Error processing your request.", str(e), None


# Gradio Interface
ui = gr.Interface(
    fn=ask_ai,
    inputs=gr.Audio(
        sources=["microphone", "upload"],
        type="filepath",
        label="Ask me a question (record or upload audio)"
    ),
    outputs=[
        gr.Textbox(label="Transcribed Question"),
        gr.Textbox(label="AI Answer"),
        gr.Audio(label="Answer Audio")
    ],
    title="🎤 Voice Q&A with Groq AI",
    description="Record or upload an audio file, get an AI-generated spoken answer.",
)

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