Skip to content

v0.3.0 — Voice-to-text dictation

Choose a tag to compare

View all tags
@oxyc oxyc released this 28 May 00:13
· 24 commits to main since this release
v0.3.0
23e6c93

Voice-to-text dictation

A mic button in the composer dictates speech into the input via the browser's Web Speech API — no backend, no key, no cost. Hidden where unsupported (e.g. Firefox); solid-red pulsing state while recording.

  • Language picker: a small chevron toggle by the mic opens a menu of the site's languages (sourced from Polylang, with the BCP-47 codes Web Speech needs), marking the selected one — so you can dictate in FI/SV even on an English admin. Remembers the last choice.
  • Send button is revealed only once the composer has text (pops in).
  • Backend-agnostic useVoiceInput hook so a server-side Whisper transcriber can drop in later.

e2e: dictation (mocked Web Speech), language selection drives recognition.lang, send-reveal, and graceful hiding when unsupported.

Assets 3
Loading