metadata

title: YoutubeTranscriptTool
emoji: 💬
colorFrom: green
colorTo: pink
sdk: gradio
sdk_version: 5.12.0
app_file: app.py
pinned: false
license: apache-2.0
short_description: Transcribe Youtube Video

YouTube Transcript SmolAgent Tool

Overview

This is a tool to fetch transcripts for YouTube videos. It attempts to retrieve the official YouTube transcript first, and if that fails or if the user opts to skip it, it falls back to using Whisper ASR to transcribe the audio.

Project Structure

app.py: Main application file that sets up the Gradio interface and handles the transcript fetching logic.
transcription.py: Contains functions for downloading audio, converting it to WAV format, and transcribing it using Whisper ASR.
youtube_utils.py: Contains utility functions for extracting video IDs and fetching official YouTube transcripts.
logging_config.py: Configures logging for the application.
ffmpeg_setup.py: Ensures that ffmpeg is available in the system path.

Dependencies

Python
pip
Gradio
yt-dlp
torch
transformers
youtube_transcript_api
ffmpeg

How It Works

User Input: The user provides a YouTube URL and optionally chooses to skip the official transcript check.
Transcript Fetching:
- If the user opts to skip the official transcript, the tool directly uses Whisper ASR to transcribe the audio.
- Otherwise, it first attempts to fetch the official YouTube transcript.
- If the official transcript is not found, it falls back to using Whisper ASR.
Whisper ASR:
- Downloads the best audio track using yt-dlp.
- Converts the audio to WAV format using ffmpeg.
- Transcribes the audio using the Whisper ASR model from the transformers library.
Output: The transcript and logs are displayed in the Gradio interface.

Running the Application

Ensure you have all dependencies installed:

pip install gradio yt-dlp torch transformers youtube_transcript_api

Run the application:
```
python app.py
```
Open the provided URL in your browser to access the Gradio interface.