CaptionitStart free

Effective 30 September 2026

Footage credits

The talking-head clips on the home page come from Wikimedia Commons under the licences below. Each was trimmed, cropped, transcribed and captioned with Captionit; the captions are our addition and are shared under the same licence as the clip.

Home page clips

Bundled sound effects

The starter pack's files are CC0 (public domain) works by the authors below, recorded per file at ingestion. The meme sounds are our own original synthesized effects — generated in code in the style of the classic sounds, never copies of the famous recordings — and are likewise dedicated CC0. CC0 needs no attribution — these credits are kept so the record stays checkable. The same record travels with each copy added to a project's media bin.

Translation models

Caption translation runs on-device in your browser with OPUS-MT models by Helsinki-NLP (Language Technology Research Group, University of Helsinki), converted to ONNX for Transformers.js. Nothing you translate is uploaded. Each model below is credited under its licence; the German model is CC-BY-4.0 and is used with attribution to Helsinki-NLP as its licence requires.

Face detection model

The clipper’s optional face-centring runs on-device in your browser with YuNet (face_detection_yunet_2023mar) from the OpenCV Zoo · MIT licence · © Shiqi Yu · about 0.23 MB, loaded from this site only when you ask it to centre a crop window. It detects faces only — no recognition, nothing about the frames leaves your device, and no detection result is stored: the outcome is a crop position you can change, exactly like one you set by dragging. Inference uses ONNX Runtime Web (MIT), the same runtime the speech and text models already use.

Questions about this document? Write to work.techhacksai@gmail.com. Parvez Achhan Ali Siddiqui, an individual (sole proprietor) trading as Captionit, India.