Ask & listen (speech)

Plays a prompt and listens for the caller’s spoken reply, then branches on what they said. The transcript lands in vars.recognizedText (confidence in vars…

What it does

Plays a prompt, then listens for the caller's spoken reply and branches on what they said — the voice equivalent of a "press 1 for sales" menu. The recognized transcript lands in vars.recognizedText (and the recognizer's confidence, when the vendor provides it, in vars.speechConfidence).

It uses the platform's default speech recognizer (configured under Admin → Speech recognition). Set a per-node Language override or Expected phrases (hints) to bias recognition toward the words you expect.

Two outgoing edges: call.speech.ended fires when a transcript was heard, call.speech.timeout when nothing was said before the listen window expired. The node also supports cases on itself, so a one-of-many voice menu can branch directly — e.g. contains "sales" → Sales.

When to use it

• A spoken IVR menu: "Say sales, support, or billing" → branch on keywords
• Yes/no confirmation by voice
• Anywhere a DTMF menu (gatherDigits) would work but you want it hands-free

Notes

• Requires a recognizer configured under Admin → Speech recognition; with none configured the listen falls back to the Jambonz account-default vendor.
• STT usage is metered per audio minute against the recognizer's rate.