Voice following uses the browser’s Web Speech API. It is most reliable in desktop Chrome; Safari support varies by device. Other browsers may expose the API without a working service, and Firefox generally requires steady pace or manual scrolling. Voice recognition may require an internet connection and may send audio to the browser provider. Browser support details.

Camera recording requires HTTPS, camera and microphone permission, and MediaRecorder support. If permission fails, choose practice without recording. Steady pace and manual scrolling work without speech recognition. In manual mode, use arrow keys, the position slider, or tap a word to resume there.

Watch the reference first, then practice with headphones. Compare vowel sounds, stress, rhythm and pauses. These are listening exercises; no automated accent score is calculated. Historical speeches contain period language and viewpoints. Accent labels describe the reference performance, not a standard you must adopt.

No recording is uploaded to Accent Prompter. Text and recordings are held in this tab, not saved on our servers. Reference players load only when requested and connect to YouTube. Closing or reloading loses your take. Sessions stop automatically at 15 minutes to limit memory use.