Listen
Waits for the user to say something and transcribes it to text with the phone’s own speech recogniser, needing no AI connection or API key.
An action in the Ask the User group.
Its type is action.listen.
Declared ports
Section titled “Declared ports”| Port | Label | Flow | Type |
|---|---|---|---|
in |
in | execution in | — |
confirmed |
When something was said | execution out | — |
cancelled |
When nothing was said | execution out | — |
timed_out |
When time runs out | execution out | — |
text |
Heard | data out | String |
Configuration
Section titled “Configuration”| Field | Key | Default | Notes |
|---|---|---|---|
| Language | language |
— | chosen from RECOGNITION_LANGUAGE |
| Give up after (seconds) | maxSeconds |
15 |
— |
| Stop after this much silence | silenceSeconds |
0 |
seconds, 0 = let the phone decide |
| Recognise on the device where possible | preferOffline |
true |
— |
- microphone access
- hardware: speech recognition
A missing permission is flagged in the editor rather than blocking the macro — the graph is fine, the phone is not, and granting it starts the node working with no edit here at all.