video1.0<iframe src="https://www.loom.com/embed/99af0738c8194dc0bf3946bfac542b36" frameborder="0" width="3342" height="2506" webkitallowfullscreen mozallowfullscreen allowfullscreen></iframe>25063342Loomhttps://www.loom.com25063342https://cdn.loom.com/sessions/thumbnails/99af0738c8194dc0bf3946bfac542b36-61fbc3234998f003.gif967.834This Loom explains how to set up voice to voice communication between an AI agent and the user. It first shows a practical method in Telegram where the agent receives a voice note and replies with a voice note, using Microsoft Edge TTS with a British male voice, and recommends keeping voice notes about 30 to 60 seconds. It then demonstrates voice to voice in Azaris, including enabling a Jarvis mode sidebar, selecting voices and speed, and adding an 11Labs API key to clone a custom “Jarvis” voice using an 11Labs voice ID and V3 for natural sound. The Loom emphasizes that while voice to voice is cool, sending back short voice notes is more practical than longer spoken responses, and the real value is automating business workflows.