Dictation API by AssemblyAI - Add fast, accurate dictation with a single line API call

Excited to be back on Product Hunt announcing the introduction of our 🆕 Dictation API.

Every product team we talk to is adding dictation right now: notes apps, clinical tools, coding agents, booking flows. And almost all of them hit the same problem, Speech-to-text gives you every word the person said, filler and false starts included. What dictation needs is the exact text the person meant, ready to hit send.

That is exactly what the Dictation API was built for. With one call, you get finished text back, corrections fixed, stutters removed -"ums" and "ahs" and all. In our benchmarks, it has been lightening fast: short clips returned in under one second (p50 at 134ms). It runs on our flagship Universal-3.5 Pro, covers 19 languages, and is fully customizable via prompting.

I built the live demo into our product page, and the thing that sold me was watching the verbatim transcript and the finished text land side by side, with the response time printed right there beside it, all included in the API's response in under a second.

If you'd rather try it yourself, our team built and released Blurt today alongside the launch. Blurt is a free, open-source macOS dictation app designed for you to hold a key, talk and get finished text right where your cursor is.

Our team will be here all day in the comments, and we're running a Dictation Hackathon in our Discord right now. So join us to chat about what you're building dictation into, get help with whats breaking, and let us know what you think!