NEW AI Voice Agents for Incredible PBX

[drmurdoch]

Anyone use one ?
They seem indispensable and I've just started trying them.

Asterisk AI Voice Agent: Open Source AI Calls on Your Own PBX

OPTION 1​

GitHub - ictinnovations/asterisk-ai-voice-agent: Self-hosted AI voice agent for Asterisk. Streaming STT to LLM to TTS over AudioSocket, with barge-in and tool calling.

Self-hosted AI voice agent for Asterisk. Streaming STT to LLM to TTS over AudioSocket, with barge-in and tool calling. - ictinnovations/asterisk-ai-voice-agent

github.com

 

[drmurdoch]

OPTION 2:​

GitHub - hkjarral/AVA-AI-Voice-Agent-for-Asterisk: An open-source AI Voice Agent that integrates with Asterisk/FreePBX using Audiosocket/RTP/Websocket technology https://demo.agent6789.com

An open-source AI Voice Agent that integrates with Asterisk/FreePBX using Audiosocket/RTP/Websocket technology https://demo.agent6789.com - hkjarral/AVA-AI-Voice-Agent-for-Asterisk

github.com

Demo HomePage : https://demo.agent6789.com/

Try the US demo lines and follow the current voice menu:

The voice lines are there to answer questions about AVA-AI-Voice-Agent-for-Asterisk

Standard voice: (925) 736-6718 (works, they get you to choose the AI behind the scenes, and they text you a summary of the conversation. WOW.
HD voice: (909) 788-2282 (didn't try it).

 

[KNERD]

The problem with using cloud services is having massive delay in responses. You would have to run one locally to get instant responses.

I have something i have been working on using only an old RTX 2070 8GB and a 12 Core Xeon. The SST runs on the CPU, A small LLM of only 4B parameters on the GPU with TTS taking up the rest of the vRAM on the same GPU. It responds instantly

 

[KNERD]

Here is a screenshot

 

[drmurdoch]

cloud services

I assumed they can be run anywhere (cloud, local)

OPTION 1:

GitHub - ictinnovations/asterisk-ai-voice-agent: Self-hosted AI voice agent for Asterisk. Streaming STT to LLM to TTS over AudioSocket, with barge-in and tool calling.

Self-hosted AI voice agent for Asterisk. Streaming STT to LLM to TTS over AudioSocket, with barge-in and tool calling. - ictinnovations/asterisk-ai-voice-agent

github.com

The brain is Anthropic Claude, streamed token-by-token so the agent starts replying before the whole answer is ready. Swapping in another LLM means implementing one small module interface, documented in PORTING.md.

You could use a local LLM ?

I had good response times in the free call demo.

OPTION 2:

GitHub - hkjarral/AVA-AI-Voice-Agent-for-Asterisk: An open-source AI Voice Agent that integrates with Asterisk/FreePBX using Audiosocket/RTP/Websocket technology https://demo.agent6789.com

An open-source AI Voice Agent that integrates with Asterisk/FreePBX using Audiosocket/RTP/Websocket technology https://demo.agent6789.com - hkjarral/AVA-AI-Voice-Agent-for-Asterisk

github.com

284 Forks.
seems like it can be run locally.

 

[KNERD]

Mine is currently set up to use local, OpenAI, Claude, and Groq.