Integrate my voice enabled AI Agent with Asterisk

[jriggin]

Greetings!
I have a voice enabled AI Agent that answers questions about Medicare. I want to connect it as an asterisk extension (or IVR or AGI or AMI or dial plan) but haven’t a clue about how to do this. I have all the Asterisk skills to do this, but it seems something like a voice enabled RAG would be the best option. Using a combination of TTS/STT works but the lag between question and answer is too long. The agent works fast and reliable over the computer speaker and mic.

Can anyone give me some guidance or point me in the right direction to get started.?

[jcolp]

Asterisk Server Configurations for Real-Time External Media Listening:

[stream-audio] ; Handle calls between any two extensions in the 7000 range using external media exten => _7XXX,1,NoOp(Starting call to ${EXTEN} from ${CALLERID(num)}) same => n,ExternalMedia(rtp,3.252.149.9:3000/3.252.149.9:3001,slin16) ; Send RTP to your Node.js server ports same => n,Hangup() ; Voicemail handling for any extensions in the 8000 range exten => _8XXX,1,VoicemailMain(${EXTEN}@main) same => n,Hangup()

Is it possible to transcribe live calls with asterisk?

All of the components are present to be able to do so. It’s up to you to put the pieces together. ARI has the ability to snoop on a channel for example[1] which can be used to passively fork audio. Once done ARI also provides external media[2] functionality to direct media outside of Asterisk to an application. What happens there is up to you. There is an example of an external media implementation on Github[3]. If you’re wanting a complete out of the box solution though, then that doesn’t exist…

There’s lots’o’threads. This topic seems to come up once or twice a week at this point in one form or another. The gist is that ARI has the pieces, but it’s up to you to use them and build something.

(There are also other options that some people use and can perhaps comment on, but I can 100% say that ARI can do this)