Skip to main content

Using voice input

How voice input works, when to use it, and tips for clean transcription.

A
Written by Andres canella
Updated today

Tap the mic icon on the input bar, speak your prompt, and stop. The audio is transcribed and sent as a routine request.

When voice is better than text

  • Long or nuanced prompts — easier to say than type

  • While moving — changing, walking to the gym, between sets

  • Describing how you feel"My hamstrings are sore from yesterday and I want something gentle that still moves."

Tips for clean transcription

  • Speak naturally, not slowly — normal pace works best

  • Quiet environments help; background noise is the usual culprit when transcription is off

  • Short pauses between clauses are fine; long silences end the recording

  • You don't need to dictate punctuation

What if the transcription is wrong

If the transcribed prompt misses something important, you can edit the routine after it generates — see Editing a routine by talking to it. Or tap the mic again and re-record.

Size limits

Voice prompts are capped around 9 MB, which is roughly a couple of minutes of speech. In practice, you'll never hit that — good prompts are usually under 20 seconds.

Privacy

Audio is sent to our server for transcription, then discarded. We don't keep the recording — only the transcribed text, which becomes part of your routine's prompt history. See What data we collect and why.

Did this answer your question?