yt-transcripts 🎼

A Python CLI tool for extracting transcripts from YouTube videos, playlists, and channels.

Installation

pip install -e .

Or install dependencies directly:

pip install youtube-transcript-api yt-dlp

With AI Summarization

To enable AI-powered summarization:

pip install -e ".[summarize]"

Usage

yt-transcripts [OPTIONS] SOURCE...

Sources

The tool accepts multiple source types:

• Video URL: https://www.youtube.com/watch?v=VIDEO_ID
• Video ID: dQw4w9WgXcQ
• Channel URL: https://www.youtube.com/@ChannelName
• Playlist URL: https://www.youtube.com/playlist?list=PLAYLIST_ID

Options

Option	Description
-f, --format	Output format: text, json, srt, vtt (default: text)
-l, --language	Preferred language code(s), can be specified multiple times (default: en)
-o, --output	Output file or directory (default: stdout)
--max-videos	Maximum number of videos to process from channel/playlist
--list-only	Only list videos without extracting transcripts
-v, --verbose	Verbose output
-h, --help	Show help message
-s, --summarize	Summarize transcripts using AI
--model	LiteLLM model string (default: ollama/llama3.2)
--api-key	API key for cloud providers
--ollama-host	Ollama server URL (default: http://localhost:11434)

Examples

Single Video

# By URL
yt-transcripts "https://www.youtube.com/watch?v=dQw4w9WgXcQ"

# By video ID
yt-transcripts dQw4w9WgXcQ

Multiple Videos

yt-transcripts VIDEO_ID1 VIDEO_ID2 VIDEO_ID3

Output Formats

# Plain text (default)
yt-transcripts VIDEO_ID -f text

# JSON with timestamps and metadata
yt-transcripts VIDEO_ID -f json

# SRT subtitles
yt-transcripts VIDEO_ID -f srt

# WebVTT subtitles
yt-transcripts VIDEO_ID -f vtt

Save to File

# Single file
yt-transcripts VIDEO_ID -o transcript.txt

# Multiple videos to separate files in a directory
yt-transcripts VIDEO_ID1 VIDEO_ID2 -o ./transcripts/

Channels

# List all videos from a channel
yt-transcripts "https://www.youtube.com/@anthropic-ai" --list-only

# Extract transcripts from first 10 videos
yt-transcripts "https://www.youtube.com/@anthropic-ai" --max-videos 10

# Save channel transcripts to directory as JSON
yt-transcripts "https://www.youtube.com/@anthropic-ai" --max-videos 5 -f json -o ./transcripts/

Playlists

# List videos in a playlist
yt-transcripts "https://www.youtube.com/playlist?list=PLrAXtmErZgOeiKm4sgNOknGvNjby9efdf" --list-only

# Extract all transcripts from playlist
yt-transcripts "https://www.youtube.com/playlist?list=PLrAXtmErZgOeiKm4sgNOknGvNjby9efdf"

Language Selection

# Prefer Spanish, fall back to English
yt-transcripts VIDEO_ID -l es -l en

# Prefer French
yt-transcripts VIDEO_ID -l fr

AI Summarization

Summarize transcripts using LLMs. Supports Ollama (local), OpenAI, Anthropic, Gemini, and OpenRouter.

# Using local Ollama (default)
yt-transcripts -s VIDEO_ID

# Specify a model
yt-transcripts -s --model openai/gpt-4o-mini VIDEO_ID

# With API key
yt-transcripts -s --model anthropic/claude-sonnet-4-20250514 --api-key sk-ant-... VIDEO_ID

# Summarize multiple videos to a directory
yt-transcripts -s -o ./summaries/ VIDEO_ID1 VIDEO_ID2

# Summarize a playlist
yt-transcripts -s --max-videos 5 "https://www.youtube.com/playlist?list=PLAYLIST_ID"

Environment Variables

Variable	Description	Default
YT_SUMMARIZE_MODEL	Default LiteLLM model	ollama/llama3.2
OLLAMA_HOST	Ollama server URL	http://localhost:11434
OPENAI_API_KEY	OpenAI API key	-
ANTHROPIC_API_KEY	Anthropic API key	-
GEMINI_API_KEY	Google Gemini API key	-
OPENROUTER_API_KEY	OpenRouter API key	-

You can also use a .env file in your project directory.

Supported Models

• Ollama (local): ollama/llama3.2, ollama/mistral, etc.
• OpenAI: openai/gpt-4o, openai/gpt-4o-mini
• Anthropic: anthropic/claude-sonnet-4-20250514, anthropic/claude-haiku
• Gemini: gemini/gemini-1.5-flash, gemini/gemini-1.5-pro
• OpenRouter: openrouter/meta-llama/llama-3-8b-instruct

Output Formats

Text

Plain text with all segments joined together:

We're no strangers to love You know the rules and so do I...

JSON

Structured data with metadata and timestamps:

{
  "video_id": "dQw4w9WgXcQ",
  "language": "en",
  "is_generated": false,
  "segments": [
    {
      "text": "We're no strangers to love",
      "start": 18.64,
      "duration": 3.24
    }
  ]
}

SRT

Standard subtitle format:

1
00:00:18,640 --> 00:00:21,880
We're no strangers to love

2
00:00:22,640 --> 00:00:26,960
You know the rules and so do I

VTT

WebVTT subtitle format:

WEBVTT

00:00:18.640 --> 00:00:21.880
We're no strangers to love

00:00:22.640 --> 00:00:26.960
You know the rules and so do I

Error Handling

The tool gracefully handles common errors:

• Transcripts disabled: Reports when a video has transcripts turned off
• Video unavailable: Reports when a video is private or deleted
• No transcript found: Reports when no transcript exists in the requested language

Errors are included in the output rather than stopping execution, so batch processing continues even if some videos fail.

Dependencies

Core:

• youtube-transcript-api - Transcript extraction
• yt-dlp - Channel and playlist video listing

Summarization (optional):

• litellm - Unified LLM interface
• python-dotenv - Environment file loading

License

MIT