YouTube Channel Transcript Extractor

Extract transcripts from an entire YouTube channel in one command, and export to four formats ready for analysis, search, or AI pipelines.

v14.1 - Production-ready single-file script. Works in Google Colab and locally.

Features

• Bulk extraction - transcripts from hundreds of videos at once
• Auto language detection - Hindi, English, and more (via langdetect)
• Shorts + long videos - filter or grab both
• Two transcript sources - YouTube Transcript API first, yt-dlp fallback
• Async + thread-safe - configurable workers, cached, rate-limited
• Anti-block - tv_embedded client + smart sleep
• 4 output formats: Excel, SQLite (FTS5), TSV, JSONL (LLM-ready)
• Google Drive - optional save to MyDrive/YT_Transcripts/
• ZIP auto-download - when Drive is not connected (Colab)
• Ctrl+C safe - partial results always saved
• LLM-optimized - trim_for_llm() keeps head+tail per transcript

Quick Start (Colab)

1. Open Google Colab
2. Paste yt_transcript_extractor.py into a cell or upload it
3. Run !python yt_transcript_extractor.py
4. Follow the prompts

Local Install

Prerequisites: Python 3.9+ and ffmpeg.

git clone https://github.com/USERNAME/yt-transcript-extractor.git
cd yt-transcript-extractor
pip install -r requirements.txt
python yt_transcript_extractor.py

Output Formats

• Excel - filter, sort, hyperlinks, frozen header. Transcripts over 32,000 chars split across columns.
• SQLite - FTS5 full-text search:

  SELECT title FROM videos_fts WHERE videos_fts MATCH 'quantum';

• TSV - UTF-8 with BOM, opens cleanly in Sheets/Excel.
• JSONL - one object per line:

  {"t":"title","v":1.2,"d":"15-01-2024","tp":"video","tr":"head [...] tail"}

Configuration

Constant	Default	Description
MAX_VIDEOS	500	Max videos per channel
WORKERS	10	Concurrent fetches
BATCH_SIZE	50	Date-fetch batch size
LLM_MAX_WORDS	500	JSONL transcript limit
RATE_LIMIT_DELAY	2.0	Seconds between channels
MAX_RETRIES	3	Retries per transcript
CHANNEL_URLS	[]	Non-interactive mode

Troubleshooting

Problem	Fix
yt-dlp not running	pip install -U yt-dlp
All transcripts N/A	Captions disabled or IP blocked - try VPN
ffmpeg not found	Install ffmpeg
Rate limited	Increase RATE_LIMIT_DELAY
asyncio error in Jupyter	pip install nest-asyncio

Legal

Personal research, accessibility, and archival use only. Respect YouTube ToS, creator copyright, and rate limits.

License

MIT - see LICENSE.