Yes, through the API. The agent needs the Web Voice Channel skill and must be Running.
Your backend calls
POST /UserAgent/Realtime/WebStartwith your project's API key. Never put the key in the browser.It returns
wsUrl,sessionToken,cancelTokenandexpiresAt. Your page opens the WebSocket at wsUrl, sends the sessionToken in its first message, then streams 24 kHz mono microphone audio.If the call never gets going, for example because the visitor blocks the microphone, call
POST /UserAgent/Realtime/Cancelwith the cancelToken.
There's an hourly session limit (HTTP 429 when reached) and a maximum session length. Call audio is paid from the Wiro AI Models project's wallet. Guide: Web Voice. Phone calls: Twilio.
