VIDEO INTO KNOWLEDGE
In developmentAI content automation
Noting
Turn video knowledge into readable notes.
A tool that transcribes YouTube audio and restructures it into articles with an outline, supporting scenes, and categories.
HOW IT COMES TOGETHER
The core workflow- 01
Video and audio
- 02
Transcription
- 03
Outline and article
- 04
Scenes and categories
01 / THE QUESTION
The problem we set out to solve.
Finding an idea again in a long video means repeatedly moving through playback. Turning it into an article also involves transcription, outlining, images, and categorization.
02 / OUR APPROACH
Our approach.
We separate transcription, outlining, and writing into stages, then link relevant timestamps to the text. The web interface and API server work together as one product.
- 01
Audio transcription
Video audio becomes text that can be used as source material for later stages.
- 02
Staged article generation
An outline is created before the body text, giving the article a clear structure.
- 03
Relevant scene links
Timestamps related to each section connect the writing with the video's visual context.
- 04
Content categorization
Generated articles are analyzed and assigned categories to make them easier to explore.
03 / INTELLIGENCE AT WORK
Where AI does the work.
Speech recognition and LLMs work together. Article generation, timestamp selection, and classification are separate tasks, with fallback handling when generation fails.
04 / WHERE WE ARE
Where the project stands.
We are developing the web interface and API server together. Transcription, article generation, scene matching, and categorization are organized as separate stages with fallback handling.
