app.py

import whisper
from groq import Groq
from gtts import gTTS
import os
import gradio as gr
import tempfile
import logging

# Setup logging
logging.basicConfig(level=logging.DEBUG)

# Initialize the Whisper model
whisper_model = whisper.load_model("base")

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

def process_audio(input_audio):
    try:
        # Log received audio file
        logging.debug(f"Received audio file: {input_audio}")

        # Transcribe audio with Whisper
        transcription = whisper_model.transcribe(input_audio)
        if "text" not in transcription or not transcription['text']:
            raise ValueError("Whisper failed to transcribe the audio.")

        user_text = transcription['text']
        logging.debug(f"Transcription: {user_text}")

        # Interact with LLM via Groq API
        response = client.chat.completions.create(
            messages=[{"role": "user", "content": user_text}],
            model="llama3-8b-8192",
        )

        if not response.choices:
            raise ValueError("Groq API returned an empty response.")

        llm_response = response.choices[0].message.content
        logging.debug(f"LLM Response: {llm_response}")

        # Convert LLM response to speech with gTTS
        if not llm_response.strip():
            raise ValueError("LLM response is empty or invalid.")

        tts = gTTS(llm_response)
        temp_output = tempfile.NamedTemporaryFile(delete=False, suffix=".mp3")
        tts.save(temp_output.name)
        logging.debug(f"Generated audio file: {temp_output.name}")

        # Return response text and audio file path
        return llm_response, temp_output.name

    except Exception as e:
        logging.error(f"Error in process_audio: {e}")
        return f"An error occurred: {str(e)}", None

# Create Gradio interface
interface = gr.Interface(
    fn=process_audio,
    inputs=gr.Audio(type="filepath"),  # Using 'filepath' to get the file path
    outputs=[
        gr.Textbox(label="LLM Response"),
        gr.Audio(label="Response Audio"),
    ],
    title="Real-Time Voice-to-Voice Chatbox",
    description="Transcribes input audio, interacts with an LLM via Groq API, and generates audio responses."
)

# Launch Gradio app
interface.launch(share=True)