AVA has been introduced—an open-source project that allows for the integration of modern artificial intelligence models into existing Asterisk and FreePBX telephony infrastructure to create intelligent voice assistants.
What Happened
A developer has introduced AVA, an open-source voice AI agent designed to work with Asterisk and FreePBX telephony systems. The project utilizes Audiosocket and RTP technologies to ensure minimal latency during audio transmission. Thanks to its modular architecture, the system allows for the combination of various STT, LLM, and TTS providers, including OpenAI Realtime, Google Live API, Grok, Deepgram, and ElevenLabs.
Context
Traditional IP telephony is often limited by rigidly programmed voice menus (IVR). The emergence of AVA creates an open bridge between classical telephony and modern machine learning models, allowing companies to move away from proprietary cloud platforms toward flexible self-hosted solutions.
Why It Matters for the Industry
The project provides an open alternative to closed commercial solutions for telephony automation. This allows companies to deploy AI operators locally or in a hybrid mode, maintaining full control over the privacy and security of audio data, which is critical for the corporate sector.
Why It Matters for Users
For users and businesses, the tool transforms a standard IP-PBX into a system with intelligent assistants. Thanks to tool calling support, agents can do more than just hold a dialogue; they can perform real tasks, such as transferring calls or sending emails, directly from the telephone interface.
What Is Not Yet Known / Limitations
The readiness of the solution for full-scale industrial operation depends on the chosen technology stack and the ability to minimize network latency when using external APIs.
Sources
Author
Look at AI, Editorial Staff