app.py

from pydub import AudioSegment
from pydub.playback import play
import tempfile
import whisper
from gtts import gTTS
from groq import Groq
import gradio as gr
import warnings
# ignore warnings
warnings.filterwarnings("ignore", category=FutureWarning)

# Load Whisper Model
model = whisper.load_model("base")  # Choose appropriate model size

# API key
GROQ_API_KEY = "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"

# Initialize Groq API Client
client = Groq(api_key=GROQ_API_KEY)

# Function for real-time voice-to-voice chatbot
def chatbot(audio_file):
    # Step 1: Transcribe audio to text
    transcription = model.transcribe(audio_file)
    user_input = transcription["text"]
    print(f"User Input: {user_input}")
    
    # Step 2: Get response from Llama model (Groq API)
    chat_completion = client.chat.completions.create(
        messages=[
            {
                "role": "user",
                "content": user_input,
            }
        ],
        model="llama3-8b-8192",
        stream=False,
    )
    llama_response = chat_completion.choices[0].message.content
    print(f"Llama Response: {llama_response}")
    
    # Step 3: Convert Llama response to speech
    tts = gTTS(llama_response)
    output_audio_path = tempfile.NamedTemporaryFile(suffix=".mp3", delete=False).name
    tts.save(output_audio_path)

    # Step 4: Process the audio (Optional for local testing)
    audio_segment = AudioSegment.from_file(output_audio_path, format="mp3")
    play(audio_segment)

    return llama_response, output_audio_path

# Gradio Interface
def interface(audio_file):
    text_response, audio_path = chatbot(audio_file)
    return text_response, audio_path

# Launch Gradio App
gr.Interface(
    fn=interface,
    inputs=gr.Audio(type="filepath", label="Speak to Chatbot"),
    outputs=[
        gr.Textbox(label="Chatbot Response"),
        gr.Audio(label="Voice Response"),
    ],
    title="Real-Time Voice-to-Voice Chatbot"
).launch()