app.py

import os
from dotenv import load_dotenv
import streamlit as st
from groq import Groq
import tempfile
import requests  # Use for calling APIs if Groq's SDK doesn't support transcription

# Load environment variables
load_dotenv()

# Initialize Groq client
client = Groq(api_key=os.getenv("GROQ_API_KEY"))

# Streamlit UI
st.title("Voice-Voice Application")
st.markdown("Listen your voice using Whisper for transcription and TTS for voice generation.")

# Upload audio file
uploaded_file = st.file_uploader(
    "Upload your audio file for transcription",
    type=["wav", "mp3", "mp4", "m4a"]
)

if uploaded_file is not None:
    # Display uploaded audio
    audio_format = uploaded_file.type.split('/')[-1]
    st.audio(uploaded_file, format=f"audio/{audio_format}")
    st.write("Processing your audio file...")

    # Save the uploaded file to a temporary location
    with tempfile.NamedTemporaryFile(delete=False, suffix='.' + audio_format) as temp_audio:
        temp_audio.write(uploaded_file.read())
        temp_audio_path = temp_audio.name

    try:
        # Call transcription API (adjust as per Groq API documentation)
        transcription_response = client.audio.transcriptions.create(
            file=open(temp_audio_path, "rb"),  # Use binary file for API
            model="whisper-large-v3-turbo",
            response_format="text"  # Adjust format if needed
        )

        # Extract the transcribed text
        transcribed_text = transcription_response  # May vary; adjust based on API response
        st.success("Transcription completed!")
        st.write("**Transcribed Text:**", transcribed_text)

        # Voice Cloning (TTS Integration)
        st.markdown("---")
        st.subheader("Generate Speech from Transcription")
        tts_input = st.text_area("Enter text to generate speech:", value=transcribed_text)

        if st.button("Generate Speech"):
            if tts_input:
                # Use a TTS system to generate audio (placeholder)
                tts_response = requests.post(
                    "https://tts.api.url",  # Replace with actual TTS API URL
                    json={"text": tts_input, "voice": "en-US-Wavenet-D"}  # Adjust parameters
                )
                if tts_response.status_code == 200:
                    st.audio(tts_response.content, format="audio/mp3")
                    st.success("Speech generation successful!")
                else:
                    st.error(f"Error in TTS: {tts_response.json()}")
            else:
                st.warning("Please enter some text.")

    except Exception as e:
        st.error(f"Error during processing: {e}")
    finally:
        # Clean up temporary file
        os.remove(temp_audio_path)

# Footer
st.markdown("Developed with ❤️ by Sanam Iftakhar")