Skip to content

video-analyzer

Describe video content locally with vision and speech models

Visit official site

What video-analyzer is

The tool is described as a video analysis utility that combines the Llama 11B vision model with OpenAI Whisper to describe what happens in a video. Key frames are extracted, fed to the vision model and joined with available transcripts to produce a natural language description.

What you can do with it

  • Extract key frames from a video intelligently
  • Transcribe audio with OpenAI Whisper
  • Analyse frames with Ollama and the Llama3.2 11B vision model
  • Generate a natural language description of the video content

How it works

  • The tool runs fully locally with no cloud service or API key
  • Frame details and available transcripts are combined
  • Computer vision, audio transcription and language processing are joined
  • An openrouter LLM service is listed for speed and scale

Getting started

  • Clone the repository from GitHub
  • Enter the project directory
  • Create a Python 3.11 virtual environment
  • Activate the virtual environment

Pros & cons

✓ What we like

  • Key frames, transcripts and frame details are combined into one description
  • The tool runs locally without a cloud service or API key

! What to watch out for

  • Descriptions depend on the extracted key frames and available transcripts
  • The documented setup uses a Python 3.11 virtual environment

FAQ

Does it need an API key?

The feature list states fully local operation with no cloud service or API key, and also mentions an openrouter LLM service for speed and scale.

Which models are used?

The material names OpenAI Whisper for transcription and Ollama with the Llama3.2 11B vision model for frame analysis.

Last reviewed: 2026-11-15

More AI video generation tools

View all →

How we review