app.py

import os
import tempfile
import numpy as np
import gradio as gr
import whisper
from gtts import gTTS
from groq import Groq
import soundfile as sf

# Set up Groq API key
os.environ['GROQ_API_KEY'] = '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'
groq_client = Groq(api_key=os.environ.get('GROQ_API_KEY'))

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

def process_audio(audio_file_path):
    try:
        # Ensure audio_file_path is valid
        if not audio_file_path:
            raise ValueError("No audio file provided")
        
        print(f"Received audio file path: {audio_file_path}")

        # Read the audio file from the file path
        with open(audio_file_path, 'rb') as f:
            audio_data = f.read()

        # Save the audio data to a temporary file
        with tempfile.NamedTemporaryFile(delete=False, suffix='.wav') as temp_audio_file:
            temp_audio_path = temp_audio_file.name
            temp_audio_file.write(audio_data)
        
        # Ensure the temporary file is properly closed before processing
        temp_audio_file.close()

        # Transcribe audio using Whisper
        result = whisper_model.transcribe(temp_audio_path)
        user_text = result['text']
        print(f"Transcribed text: {user_text}")
        
        # Generate response using Llama 8b model with Groq API
        chat_completion = groq_client.chat.completions.create(
            messages=[
                {
                    "role": "user",
                    "content": user_text,
                }
            ],
            model="llama3-8b-8192",
        )
        response_text = chat_completion.choices[0].message.content
        print(f"Response text: {response_text}")
        
        # Convert response text to speech using gTTS
        tts = gTTS(text=response_text, lang='en')
        with tempfile.NamedTemporaryFile(delete=False, suffix='.mp3') as temp_audio_file:
            response_audio_path = temp_audio_file.name
            tts.save(response_audio_path)

        # Ensure the temporary file is properly closed before returning the path
        temp_audio_file.close()
        
        return response_text, response_audio_path
    except Exception as e:
        return f"Error: {str(e)}", None

# Create Gradio interface with updated layout
with gr.Blocks() as demo:
    gr.Markdown(
        """
        <style>
        .gradio-container {
            font-family: Arial, sans-serif;
            background-color: #e0f7fa;  /* Changed background color */
            border-radius: 10px;
            padding: 20px;
            box-shadow: 0 4px 12px rgba(0,0,0,0.2);
        }
        .gradio-input, .gradio-output {
            border-radius: 6px;
            border: 1px solid #ddd;
            padding: 10px;
        }
        .gradio-button {
            background-color: #28a745;
            color: white;
            border-radius: 6px;
            border: none;
            padding: 8px 16px;  /* Adjusted padding */
            font-size: 16px;  /* Adjusted font size */
        }
        .gradio-button:hover {
            background-color: #218838;
        }
        .gradio-title {
            font-size: 24px;
            font-weight: bold;
            margin-bottom: 20px;
        }
        .gradio-description {
            font-size: 14px;
            margin-bottom: 20px;
            color: #555;
        }
        </style>
        """
    )
    
    gr.Markdown("# Voice-to-Voice Chatbot\nDeveloped by Salman Maqbool")
    gr.Markdown("Upload an audio file to interact with the voice-to-voice chatbot. The chatbot will transcribe the audio, generate a response, and provide a spoken reply.")
    
    with gr.Row():
        with gr.Column():
            audio_input = gr.Audio(type="filepath", label="Upload Audio File")
            submit_button = gr.Button("Submit")
        
        with gr.Column():
            response_text = gr.Textbox(label="Response Text", placeholder="Generated response will appear here")
            response_audio = gr.Audio(label="Response Audio", type="filepath")

    submit_button.click(process_audio, inputs=audio_input, outputs=[response_text, response_audio])

# Launch the Gradio app
demo.launch()