app.py

import streamlit as st
from faster_whisper import WhisperModel
import logging
import tempfile
import os
import time  # For simulating progress and delay

# Configure logging for debugging purposes
logging.basicConfig()
logging.getLogger("faster_whisper").setLevel(logging.DEBUG)

def format_timestamp(seconds):
    """Convert seconds to HH:MM:SS.mmm format."""
    hours = int(seconds // 3600)
    minutes = int((seconds % 3600) // 60)
    seconds_remainder = seconds % 60
    return f"{hours:02d}:{minutes:02d}:{seconds_remainder:06.3f}"

def transcribe(audio_file, model_size, progress_text, progress_bar):
    # Initialize the Whisper model based on the selected model size
    device = "cpu"  # Use "cpu" for CPU, "cuda" for GPU
    compute_type = "int8"  # Use "int8" for faster inference on both CPU and GPU
    
    model = WhisperModel(model_size, device=device, compute_type=compute_type)

    # Save the uploaded file to a temporary file
    with tempfile.NamedTemporaryFile(delete=False, suffix=os.path.splitext(audio_file.name)[1]) as tmp:
        tmp.write(audio_file.getvalue())
        tmp_path = tmp.name

    # Update progress and text
    progress_text.text("Preparing file for transcription...")
    progress_bar.progress(20)
    time.sleep(1)  # Simulate processing delay

    # Transcribe the audio file
    progress_text.text("Transcribing audio...")
    segments, _ = model.transcribe(tmp_path)
    progress_bar.progress(70)
    time.sleep(1)  # Simulate processing delay

    # Clean up the temporary file
    os.remove(tmp_path)

    # Finalize progress
    progress_text.text("Finalizing...")
    progress_bar.progress(100)
    time.sleep(0.5)  # Final step delay
    progress_text.text("Transcription complete.")

    # Format and gather transcription with enhanced timestamps
    transcription_with_timestamps = [
        f"[{format_timestamp(segment.start)} -> {format_timestamp(segment.end)}] {segment.text}"
        for segment in segments
    ]
    
    return "\n".join(transcription_with_timestamps)

# Streamlit UI components
st.title("Whisper")
st.write("For API use please visit [this space](https://huggingface.co/spaces/Lenylvt/Whisper-API)")

audio_file = st.file_uploader("🎵 Upload Audio or Video", type=['wav', 'mp3', 'ogg', 'mp4', 'avi'])
model_size = st.selectbox("📁 Model Size", ["base", "small", "medium", "large", "large-v2", "large-v3"])

if audio_file is not None and model_size is not None:
    progress_text = st.empty()  # Placeholder for dynamic text updates
    progress_bar = st.progress(0)
    transcription = transcribe(audio_file, model_size, progress_text, progress_bar)
    progress_text.empty()  # Optionally clear or leave the completion message
    st.text_area("📜 Transcription", transcription, height=300)