Build Your Own AI Voice System with Asterisk and AVR
AVR combines the power of open-source technology with enterprise-grade features, giving you complete control over your voice AI solution while maintaining flexibility and scalability.
Bring your own ASR/LLM/TTS and integrate it with AVR following a simple protocol. Full freedom to choose your preferred providers.
Users can interrupt the AI and speak at any time, creating a more natural conversational flow with advanced VAD technology.
Every component runs in its own Docker container for easy scaling and deployment. Secure and isolated architecture.
Built-in and extensible support for OpenAI/Anthropic-style function calls, enabling complex voice interactions.
AVR integrates with leading AI providers and open-source solutions
Enterprise ASR solutions
LLM, Function calling and Assistants
Enterprise TTS solutions
LLM with a lot of providers
AVR stands out by offering complete flexibility and control over your voice AI solution. Unlike proprietary platforms, you choose your preferred providers and have full access to the source code.
AVR is built with scalability and flexibility in mind
Central orchestration and business logic
Speech recognition and transcription
Language model processing
Text-to-speech conversion
SIP and telephony support
Visual configuration and management
Note: If you already have your contact center on Asterisk, you don't need to do anything, but with the Asterisk AudioSocket application, you can connect your extension directly with AVR. Asterisk version 18 or later is required as it contains the precompiled AudioSocket application.
AVR is completely free and open-source. If you find value in our project and want to support its continued development, please consider making a donation. Your contribution helps us maintain and improve the platform for everyone.
If you already have your contact center on Asterisk, you don't need to do anything! With the Asterisk AudioSocket application (available in Asterisk 18+), you can connect your existing extensions directly to AVR. No complex migration is needed.
Yes, AVR is completely free and open-source under the MIT license. You can use, modify, and distribute it freely. You only pay for the third-party ASR/LLM/TTS services you choose to integrate.
Basic knowledge of Docker and Docker Compose is essential for deployment. Familiarity with Asterisk configuration (specifically `extensions.conf` and `ari.conf`) is helpful for integration. Understanding how to obtain and configure API keys for your chosen AI services (ASR, LLM, TTS) is also required.
AVR currently has built-in support for Google, Deepgram, and ElevenLabs (ASR/TTS), OpenAI and Anthropic (LLM, Function Calling, Assistants), and OpenRouter (LLM). However, its plug-and-play architecture allows you to integrate other providers by adapting them to the simple AVR service protocol.
We welcome contributions! You can contribute code, report bugs, or suggest features on our GitHub repository. For help and community discussion, please join our Discord server or use the GitHub Discussions.
Join the growing community of developers and businesses using AVR to power their voice AI applications.