[OpenSIPS-Users] Opensips-AI Voice Connector - on existing server

[Răzvan Crainea]

Hello!

Check my answers below:

1. Sure - all you have to do is to dispatch the calls you want to get to 
an AI to the CE instance
2. Depends on your setup - if you have trunks that you can dispatch to, 
all you need to do is to setup the CE as a trunk and use it.
3. Yes - you can pass a JSON to configure the AI through the 
extra_params, or use an external web server to query during handling 
(see [1])
4. If you have the SIP IP on the server, I don't see any issues
5. Is that the only error you're getting? can you run in debug mode and 
post a few more logs? A PCAP might also be helpful.

[1] https://github.com/OpenSIPS/opensips-ai-voice-connector-ce/pull/8

Best regards,

Răzvan Crainea
OpenSIPS Core Developer / SIPhub CTO
http://www.opensips-solutions.com / https://www.siphub.com

On 1/8/25 6:07 PM, HS wrote:
> Hi.
> 
> Realtime docs suggest using the following two models - unsure it makes
> sense to change in the code or via the config file.
> 
> gpt-4o-realtime-preview-2024-12-17
> gpt-4o-mini-realtime-preview-2024-12-17
> 
> I have an Opensips 3.0 installation with RTPEngine on an EC2 instance, that
> redirects calls to a Freeswitch server in case of no answer etc. I just
> installed the AI voice connector and can't get the calls to work. So far I
> have added an "openai" user in the user table and dbaliases also. Just want
> to clarify:
> 
>     1. Is it possible to make the connector work in parallel with the
>     original Opensips install?
>     2. Do I need to modify the .cfg file in the original Opensips install?
>     3. Is it possible to have multiple configurations? For eg. have a
>     different voice for ext 5555 and another for ext 5556?
>     4. Any watchouts when working with AWS (internal/external IPs)? (The IP
>     is present on the host).
>     5. Instead of sending to Freeswitch (in case of no answer), I changed
>     the cfg to keep the call local. Getting the following error:
>     ERROR:b2b_entities:b2b_ua_server_init: failed to create new b2b server
>     instance
> 
> 
> Hi.
> 
> Realtime docs suggest using the following two models - unsure it makes 
> sense to change in the code or via the config file.
> 
> gpt-4o-realtime-preview-2024-12-17
> gpt-4o-mini-realtime-preview-2024-12-17
> 
> I have an Opensips 3.0 installation with RTPEngine on an EC2 instance, 
> that redirects calls to a Freeswitch server in case of no answer etc. I 
> just installed the AI voice connector and can't get the calls to work. 
> So far I have added an "openai" user in the user table and dbaliases 
> also. Just want to clarify:
> 
>  1. Is it possible to make the connector work in parallel with the
>     original Opensips install?
>  2. Do I need to modify the .cfg file in the original Opensips install?
>  3. Is it possible to have multiple configurations? For eg. have a
>     different voice for ext 5555 and another for ext 5556?
>  4. Any watchouts when working with AWS (internal/external IPs)? (The IP
>     is present on the host).
>  5. Instead of sending to Freeswitch (in case of no answer), I changed
>     the cfg to keep the call local. Getting the following error:
>     |ERROR:b2b_entities:b2b_ua_server_init: failed to create new b2b
>     server instance |
> 
> 
> _______________________________________________
> Users mailing list
> Users at lists.opensips.org
> http://lists.opensips.org/cgi-bin/mailman/listinfo/users