app.py

import gradio as gr
import os
from groq import Groq
import whisper
import torch
from gtts import gTTS
import IPython.display as ipd

# Set up Whisper with a smaller model or on CPU
model_name = "small"  # Use "small", "base", or "medium" for smaller models
whisper_model = whisper.load_model(model_name)
device = "cuda" if torch.cuda.is_available() else "cpu"
whisper_model.to(device)

# Set up Groq API with direct API key
client = Groq(
    api_key="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"
)

# Function to transcribe audio using Whisper
def transcribe_audio(audio_path):
    try:
        print(f"Audio file path received for transcription: {audio_path}")

        # Load the audio file directly using Whisper's built-in functionality
        result = whisper_model.transcribe(audio_path)
        print(f"Transcription result: {result}")
        return result['text']
    except Exception as e:
        print(f"Error during transcription: {e}")
        return "Error during transcription"

# Function to get response from LLaMA using Groq API
def get_llama_response(transcription):
    try:
        chat_completion = client.chat.completions.create(
            messages=[
                {
                    "role": "user",
                    "content": transcription,
                }
            ],
            model="llama3-8b-8192",
        )
        return chat_completion.choices[0].message.content
    except Exception as e:
        print(f"Error during LLaMA response generation: {e}")
        return "Error during response generation"

# Function to convert text to speech using gTTS
def text_to_speech(text):
    try:
        tts = gTTS(text)
        tts.save("response.mp3")
        return "response.mp3"
    except Exception as e:
        print(f"Error during text-to-speech conversion: {e}")
        return "Error during text-to-speech conversion"

# Main function for Gradio interface
def chatbot(audio_path):
    transcription = transcribe_audio(audio_path)
    response = get_llama_response(transcription)
    audio_path = text_to_speech(response)
    return transcription, response, audio_path

# Gradio Interface
interface = gr.Interface(
    fn=chatbot,
    inputs=gr.Audio(type="filepath"),  # Use 'filepath' to work with the path of the audio file
    outputs=["text", "text", "audio"],
    live=True,
    description="Real-time Voice-to-Voice Chatbot"
)

# Launch the interface
interface.launch()