BentoVoiceAgent

About BentoVoiceAgent

This repository demonstrates how to build a voice agent using open-source Large Language Models (LLMs), text-to-speech (TTS), and speech-to-text (STT) models. It utilizes Pipecat voice pipeline and is deployed with BentoML. The voice agent is accessible via a phone number, leveraging Twilio as the communication transport. This example can be easily extended to incorporate additional voice agent features and functionality.

The LLM and XTTS models are deployed as separate API endpoints, as outlined in the instructions below. These API endpoints are provided to the voice agent through environment variables.

Deploy the LLM and XTTS models by following the instructions provided in their respective repositories.

From the project's README

BentoVoiceAgent is an open-source project written primarily in Python, with 128 stars on GitHub. It was last updated in July 2026.