Local-first transcription with a real audio chain, system-wide input, and cloud workflows on eligible plans.
Local multi-engine transcription (Whisper-class models including Large v3 Turbo, plus optional engines where available) with formatting helpers so spoken drafts need less manual cleanup.
The audio chain, the transcription engines, and the plumbing that puts words in the field you were already typing in. Filter by the part you care about.
Local engines surface progressive partials as you speak, so the sentence is on screen before you finish it. Dedicated streaming engines are available when you need lower latency.
ShipthepricingupdatebeforeFridayandloopinthedesignreview.
Pro adds Deepgram-backed cloud transcription for broader language coverage, stronger meeting workflows, and richer long-form output.
Strip rumble and HVAC hum with an adjustable frequency cutoff before a single word reaches the model.
Smart gating with attack and release controls, so the room goes quiet the moment you do.
Normalises level across a session. Lean back, lean in, or trail off — the transcript stays even.
Level meters with peak hold give you a continuous read on input quality, so a bad take is obvious while it is still fixable.
Background reduction that lets your voice through cleanly, even when the environment refuses to cooperate.
Works in any standard text field — email, docs, chat, the terminal, a browser form. No per-app integration to maintain.
Teach it your product names, teammates, and domain jargon so the words you say most stop coming back wrong.
Choose multilingual local models, or Pro cloud engines when you need non-English dictation. Some default local packs are English-first.
Start with Free, move to Plus for unlimited local transcription, or choose Pro when you need cloud workflows.