Describe your agent in plain words, in a chat. Pick a voice, or make one that is yours: record a paragraph or describe it and tune it with meters. When you like the preview, press Generate API. We run the model, keep the memory, and speak the answer sentence by sentence, using instant library phrases when they match.
Chat with the builder. It fills an editable configuration: personality, language, voice, knowledge, FAQs, actions and memory. Five agents per account.
Each publish is a numbered version at one endpoint, with a key that works only for that agent and short-lived tokens for browsers.
You get the text and one audio link per sentence. Play them in order, and report what was actually played if the caller interrupts.
curl -X POST https://speakvora.com/api/v1/agents/ag_123/turn \
-H "x-api-key: $SPEAKVORA_KEY" -H "content-type: application/json" \
-d '{"session":"visit-42","text":"Can I move my appointment to Thursday?"}'
{ "session": "visit-42",
"reply": "Of course. What day and time is your current appointment?",
"audio_url": "https://speakvora.com/live/.../....mp3",
"source": "live", "ms": 5200 }