Is it possible to receive a call automatically and connect it to a websocket?

Check out this blog and the demo project. In that project, I use IBM Watson real-time speech-to-text, but you can change to use any other 3rd party service.

Thanks for sharing the blog. I have come up with a solution.
We already have a IVR menu set up to the number. now we want to replace it with the app. how can we do that?

“With the app”, what app is that?

the node js application that I have made following the blog. I just chandge the AI to real-time api.
we have one main company number and 3 other numbers that are connected to the ivr menu. Now I what I want is to connect all incomming calls with the application.

Oh, I still don’t know what you really want to setup. But the app is a device with phone number and that “device” is assigned to a user extension. If a call is made to your IVR, route the call to that extension then the device will receive the incoming call. Is that what you want?

Exactly. But I want to replace the IVR menu. I want to route all incoming calls to the device. But I don't know how to do it. 

The easiest way is to forget about the system IVR and let it be as such for prompts IVR.

Then create a user extension (name it e.g. AI IVR) and build your app (device) and assign the device to that user extension. Promote the user extension’s number so customers can dial that number directly instead of dialing the IVR.

If the above solution is not the logic you expect, then choose 1 of the following options:

1. Set the IVR with a prompt e.g. press 1 to connect to our virtual agent etc. And set the key 1 to route the call to the user extension with that “device”
2. Implement and subscribe for the IVR telephony session event notifications and when you detect an incoming call to that IVR, forward the call to the user extension with that “device”. This solution is complicated and requires the app to be authenticated by the main super admin user (normally the user extension 101). Check this article to learn more about the telephony session even notification.

Thanks a lot. I have managed to connect the call with open AI. but having a new issue now. I am hearing a Gittery noise. the vocie is not hearable. this might be due to audio format issue. I am using “g711_ulaw”. I am sure that realtime api can handle this format. but does ringcentral handle this format?

openAIWs.on("open", () => {

console.log(`OpenAI WebSocket connected for call ${activeCall.id}`);

// Send session update to configure the call

setTimeout(() => {

try {

const sessionUpdate = {

type: "session.update",

session: {

turn_detection: { type: "server_vad" },

input_audio_format: "g711_ulaw",

output_audio_format: "g711_ulaw",

voice: OPENAI_VOICE,

instructions: CALL_PROMPT,

modalities: ["text", "audio"],

temperature: 0.7,

},

};

console.log("Sending session update to OpenAI");

openAIWs.send(JSON.stringify(sessionUpdate));

} catch (error) {

console.error("Error sending session update:", error);

}

}, 250);

});
 

These are all the audio codecs the RingCentral soft-phone SDK supports.

https://github.com/ringcentral/ringcentral-softphone-ts?tab=readme-ov-file#audio-codec

I have added codec:"PCMU/8000".
after doing so, the audio is better. but still the audio is laggy. 
I am not sure what is happening.

I am in Bangladesh. and using a app named call.com to make the calls.
is network a issue here?