Spaces:

Lenylvt
/

Whisper

Sleeping

App Files Files Community

Whisper / app.py

Lenylvt

Update app.py

f871faa verified 4 months ago

raw history blame contribute delete

No virus

2.88 kB

	import streamlit as st
	from faster_whisper import WhisperModel
	import logging
	import tempfile
	import os

	# 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):
	# Initialize placeholders for dynamic updates
	progress_text = st.empty()
	progress_bar = st.progress(0)

	# 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)

	# Update progress and text for file preparation
	progress_text.text("Preparing file for transcription...")
	progress_bar.progress(10)

	# 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 for transcription start
	progress_text.text("Transcribing audio...")
	progress_bar.progress(30)

	# Transcribe the audio file
	segments, _ = model.transcribe(tmp_path)

	# Update progress and text after transcription
	progress_text.text("Processing transcription...")
	progress_bar.progress(70)

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

	# 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
	]

	# Finalize progress and clear text
	progress_text.text("Transcription complete.")
	progress_bar.progress(100)
	progress_text.empty() # Optionally clear the completion message

	return "\n".join(transcription_with_timestamps)

	# Example Streamlit UI setup to use transcribe function
	st.title("Whisper")
	st.write("For Remove Timestamps please visit [this Space](https://huggingface.co/spaces/Lenylvt/Whisper_Timestamps_Remover). 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:
	transcription = transcribe(audio_file, model_size)
	st.text_area("📜 Transcription", transcription, height=300)