Video Summarizer

AI-powered video transcript summarizer using Venice AI API (you can alternatively use different API that is OpenAI compatible). This tool processes video transcripts and generates both concise and detailed summaries.

To get the transcripts, use whisper-file or yt-whisper.

Installation

Prerequisites

Python 3.11 or higher
Poetry (for dependency management)
Venice AI API key

Setup

Clone the repository:

git clone https://github.com/jooray/video-summarizer
cd video-summarizer

Install dependencies using Poetry:
```
poetry install
```

Set up your Venice AI API key:

export VENICE_API_KEY="your-api-key-here"

Or add it to your shell profile (.bashrc, .zshrc, etc.):

echo 'export VENICE_API_KEY="your-api-key-here"' >> ~/.zshrc
source ~/.zshrc

Create configuration file (optional):

cp video-summarizer.json-example ~/.video-summarizer.json
# Edit the config file to customize prompts and settings

Usage

Basic Usage

# Using poetry script
poetry run video-summarizer path/to/transcript.txt

# Or using python directly
poetry run python src/main.py path/to/transcript.txt

Advanced Usage

# Use custom config file
poetry run video-summarizer transcript.txt --config /path/to/custom-config.json

# Override prompts via command line
poetry run video-summarizer transcript.txt \
  --short_prompt "Summarize this video in one sentence" \
  --detailed_prompt "Provide a bullet-point summary with timestamps" \
  --merge_prompt "Combine these summaries into a unified report"

# Combine custom config and prompt overrides
poetry run video-summarizer transcript.txt \
  --config ./my-config.json \
  --short_prompt "Custom short summary prompt"

Configuration

The tool uses a JSON configuration file to customize behavior. By default, it looks for ~/.video-summarizer.json.

Configuration Options

Option	Type	Default	Description
`prompts.short_summary`	string	Basic short summary prompt	Prompt for generating concise summaries
`prompts.detailed_summary`	string	Basic detailed summary prompt	Prompt for generating comprehensive summaries
`prompts.merge_summary`	string	Basic merge prompt	Prompt for merging multiple chunk summaries into a cohesive detailed summary
`context_size`	integer	128000	Maximum tokens per chunk for processing
`model`	string	"mistral-31-24b"	Model name to use for summarization

Example Configuration

{
  "prompts": {
    "short_summary": "Provide a concise summary of this video transcript in 2-3 sentences, highlighting the main topic and key takeaways.",
    "detailed_summary": "Analyze this video transcript and provide a comprehensive summary that includes: 1) Main topics discussed, 2) Key points and insights, 3) Important quotes or specific mentions, 4) Any actionable items or conclusions. Structure your response with clear headings and bullet points where appropriate.",
    "merge_summary": "Merge the following section summaries into a cohesive, well-structured detailed summary. Eliminate redundancy, organize information logically with clear headings, and ensure smooth transitions between sections. Maintain all important details while creating a unified narrative."
  },
  "context_size": 128000,
  "model": "mistral-31-24b"
}

Command Line Options

Usage: video-summarizer [OPTIONS] TRANSCRIPT_PATH

  Summarize video transcripts using Venice API (or another OpenAI compatible API).

Arguments:
  TRANSCRIPT_PATH  Path to the video transcript file  [required]

Options:
  --short_prompt TEXT      Custom prompt for short summary
  --detailed_prompt TEXT   Custom prompt for detailed summary
  --merge_prompt TEXT      Custom prompt for merging chunk summaries
  --config PATH            Path to custom configuration file
  --help                   Show this message and exit.

Support and value4value

If you like this project, I would appreciate if you contributed time, talent or treasure.

Time and talent can be used in testing it out, fixing bugs, spreading the word.

Treasure can be sent back through here.

License

This project is licensed under the GLWTSPL