Skip to main content

Annotate anything

AfriAnnotate is built on the Label Studio engine, so the entire annotation feature set is here: text, audio, images, video, and time series; NER, classification, transcription, OCR, boxes and polygons, audio segmentation, and the rest — driven by a composable XML labeling config and a template gallery that includes a voice-collection family built for this platform's speech work.

What this fork adds on top:

  • Presentable by default — a bare config renders styled (boxed prompts, bordered options, proper input rhythm) with zero-specificity CSS, so your own <Style> always wins when you want it to.
  • The <AudioRecord> tag — in-browser speech recording with live metering and an automated QC pipeline behind it.
  • A UI your community localizes live — reviewed translations publish as live locales with no rebuild; projects can pin their annotator-facing language.

Pages in this section walk the full path: set up a project, shape the labeling interface, bring data in, and run the annotation floor — including ML-assisted pre-annotation and webhooks.