Every dictation tool has to answer one question: how do you tell it you’re talking to it? There are two answers and they feel completely different in use.
Toggle
Press a key to start, press again to stop. Apple’s dictation works this way (double-tap fn). It’s fine for long passages. The problem is the stop: you finish your sentence, forget to press, and the next thing you say to a colleague ends up in the document. Or you press once too many and lose the recording. Every toggle user has a story.
Hold-to-talk
Hold a key while you speak, let go when you’re done. Like a walkie-talkie. Speaky Box, Wispr Flow and Lemon all work this way, and it’s the right default for a reason: the key state is the recording state. You can’t forget to stop, because you’d have to keep holding. Talk for three seconds, let go, the text appears. Talk for two minutes, same thing.
It also changes how you write. Because starting costs nothing, you use it for a five-word Slack reply, not just for the long email. That’s where the time saving actually comes from.
Which key
Something you don’t use for anything else, easy to hold with one finger while your other hand is on the mouse. Right option, caps lock and fn are the popular choices. Speaky Box lets you pick any key for either of its two functions, and you can turn one off if you only want the other.
When a toggle makes sense
Long dictation with your hands busy: reading from notes, drawing, walking around with a headset. That’s a niche, and most tools with a toggle also offer hold-to-talk. Start with hold-to-talk and you’ll rarely reach for the other.