Speech to Text

Dictation is faster than typing, and different

Most people speak at 120 to 150 words per minute and type at 40 to 60, so dictation is nominally two to three times faster — and yet many people who try it abandon it quickly. The reason is that dictating is a genuinely different cognitive activity from writing. Typing gives you constant opportunity to pause, reread, reconsider and revise, and that editing loop is woven into how most people compose. Speaking is linear and unforgiving; you must hold the shape of the sentence in your head before you begin it. The people who dictate well have learned to separate the two phases — get the words out fast and messy, then edit ruthlessly afterwards — rather than trying to compose polished prose in real time.

The privacy line worth knowing

There is an asymmetry in browser speech APIs that is not widely understood and genuinely matters. Speech synthesis — turning text into audio — runs locally on your device, so the text you have it read never leaves your machine. Speech recognition typically does not: in most browsers, the audio is sent to a cloud service for processing, because accurate recognition needs models far larger than a browser wants to ship. So dictating is not a private act in the way that having text read aloud is. Treat anything you dictate as though you had typed it into a search engine, and keep confidential material — medical details, legal matters, credentials — off the microphone.