Cilevi: Windows screen recordings with automatic zooms and local captions

Cilevi is a Windows screen recorder that prepares several editing elements as part of capture. Its website describes automatic zooms based on clicks, a smoothed pointer, backgrounds and a timeline for refining the result. It also advertises speech recognition on the user’s computer for captions.

Recording and editing stay connected

The Cilevi website lists display, window and area capture, with webcam and audio tracks alongside the screen recording. The editing workspace includes cuts, transitions, annotations and blur or mosaic regions.

Click information supplies the initial zoom placement, but users can move or retime zoom clips and add their own. That makes the automatic treatment a starting point to inspect. A recording intended to teach a task still needs the right framing and enough context around the control being demonstrated.

Local captions need an accuracy check

Cilevi says caption recognition runs locally, with word-level timing and support for 99 languages. It also describes cutting a video by deleting transcript text.

Those features connect the transcript with the edit, so a recognition error can matter beyond the visible caption. A useful first trial is a short recording containing the product names and terminology the final video will use. Check the transcript before deleting material through it, and inspect any blurred regions throughout the exported clip.

Separate the released recorder from planned features

The site lists Windows 10 and 11 availability, with a Mac version awaiting review. It describes an AI workflow that plans and films a web-app tutorial as a future Version 3 Plus feature running in test builds, rather than part of the released recorder’s general offering.

Cilevi’s public description is most useful when those stages remain clear. Someone looking for automatic zooms and editable screen capture can evaluate the current recorder. Someone expecting an agent to operate an app and create a complete tutorial should first check the actual release status and supported workflows.