app.py

import os
os.environ["GROQ_API_KEY"] = "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"

import os
import gradio as gr
import whisper
from gtts import gTTS
import io
from groq import Groq

client = Groq(api_key=os.environ.get("GROQ_API_KEY"))

model = whisper.load_model("base")

def process_audio(file_path):
    try:
        audio = whisper.load_audio(file_path)

        result = model.transcribe(audio)
        text = result["text"]

        chat_completion = client.chat.completions.create(
            messages=[{"role": "user", "content": text}],
            model="llama3-8b-8192",  
        )

        response_message = chat_completion.choices[0].message.content.strip()

        tts = gTTS(response_message)
        response_audio_io = io.BytesIO()
        tts.write_to_fp(response_audio_io)  # Save the audio to the BytesIO object
        response_audio_io.seek(0)

        with open("response.mp3", "wb") as audio_file:
            audio_file.write(response_audio_io.getvalue())

        return response_message, "response.mp3"

    except Exception as e:
        return f"An error occurred: {e}", None

title = " Voice-to-Voice Chatbot Application"
description = "Developed by [Adnan Tariq](https://www.linkedin.com/in/adnaantariq/) with ❤️"
article = "### Instructions\n1. Upload an audio file.\n2. Wait for the transcription.\n3. Listen to the chatbot's response."


iface = gr.Interface(
    fn=process_audio,
    inputs=gr.Audio(type="filepath"),  # Use type="filepath"
    outputs=[gr.Textbox(label="Response Text"), gr.Audio(label="Response Audio")],
    live=True,
    title=title,
    description=description,
    theme="dark",
    article=article
)

iface.launch()