What is a wake word?
Why wake words exist
An assistant that processed everything it heard would be unacceptable for both privacy and accuracy. The wake word acts as a spoken push-to-talk: a low-power local detector watches for one phrase and gates everything else behind it, which is how always-available assistants avoid being always-recording services.
Wake words on the desktop
Phones and smart speakers made wake words familiar, but desktop voice tools often prefer a hotkey instead: at a keyboard, a key press is faster and unambiguous. Hands-free desktop modes may use a wake word, open listening with visible state, or voice activity detection, depending on the product.
Common failure modes
False accepts (the assistant wakes to a similar-sounding phrase or the TV) and false rejects (it misses you in noise) are the classic trade-off pair. Vendors tune sensitivity between them, which is why wake words are multi-syllable and phonetically distinctive.
Frequently asked questions
Is a wake-word device recording everything?
The wake-word detector runs continuously by design, but mainstream implementations process that audio locally in a short rolling buffer and transmit or store audio only after the wake word triggers. Product documentation and OS indicators are the place to verify specifics.
Do I need a wake word on a computer?
Usually not: a hotkey is faster at a keyboard, which is why desktop tools like VoiceOS center on press-to-talk with hands-free modes for when your hands are unavailable.
Keep exploring
Hear it in practice
VoiceOS puts dictation, editing, and voice-to-action in one layer on Mac and Windows.
Try VoiceOS