Sidebar — Real-Time Speaker Practice

Open the Sidebar classroom preview

No account required. Includes Jury Charge, Literary with a separate reader, a practice timer, and 1-, 2-, 3-, 4-, and 5-person practice, editable speaker titles, a portrait library, customizable keys, illustrated instructions, optional hand and voice controls, and transcript/pedal controls. Voice match is opt-in, runs the bundled local WeSpeaker speaker-embedding model in the browser, and stores profiles only in the local browser; MFCC is retained only as a model-load fallback. Forget all voices removes local profiles; uncertain matches keep the current speaker. Tone remains the default voice mode. Close your fist to reset the hand cue, then show the next number. In combined mode, lower your hand to resume voice switching.

Import PDF, Word DOCX, or text; review speaker labels; then follow the transcript with local English speech recognition. Documents and microphone audio stay on your device. The first microphone setup downloads about 40 MB. Scanned PDFs need OCR first. Recognition is experimental; Next / F8 remains available for corrections. Optional Audio practice reads with selected local device voices or plays a teacher recording with marked speaker timestamps. Voices and accents vary by device. Audio files and cue times last for the current session.

This repository contains the compiled website distribution. Authoring source is maintained privately. Browser-delivered JavaScript and assets are inspectable. Third-party licenses are provided in THIRD_PARTY_NOTICES.txt and the mediapipe, speech, and pdfjs folders.