app.py

import torchaudio as ta
import streamlit as st

from io import BytesIO
from transformers import AutoProcessor, SeamlessM4TModel

processor = AutoProcessor.from_pretrained("facebook/hf-seamless-m4t-medium", use_fast=False)
model = SeamlessM4TModel.from_pretrained("facebook/hf-seamless-m4t-medium")

# Title of the app
st.title("Audio Player with Live Transcription")

# Sidebar for file uploader and submit button
st.sidebar.header("Upload Audio Files")
uploaded_files = st.sidebar.file_uploader("Choose audio files", type=["mp3", "wav"], accept_multiple_files=True)
submit_button = st.sidebar.button("Submit")


# def transcribe_audio(audio_data):
#     recognizer = sr.Recognizer()
#     with sr.AudioFile(audio_data) as source:
#         audio = recognizer.record(source)
#     try:
#         # Transcribe the audio using Google Web Speech API
#         transcription = recognizer.recognize_google(audio)
#         return transcription
#     except sr.UnknownValueError:
#         return "Unable to transcribe the audio."
#     except sr.RequestError as e:
#         return f"Could not request results; {e}"


if submit_button and uploaded_files:
    st.write("Files uploaded successfully!")

    for uploaded_file in uploaded_files:
        # Display file name and audio player
        print(uploaded_file)
        st.write(f"**File name**: {uploaded_file.name}")
        st.audio(uploaded_file, format=uploaded_file.type)

        # Transcription section
        st.write("**Transcription**:")

        # Read the uploaded file data
        waveform, sampling_rate = ta.load(uploaded_file.getvalue())

        # Run transcription function and display
        # import pdb;pdb.set_trace()
        # st.write(audio_data.getvalue())