AI Streaming

The stream() method in the PiopiyAction class sets up a WebSocket connection to transmit the call's audio in real time. This is ideal for scenarios such as real-time AI processing, interactive voice response (IVR) systems, or any application where live call audio needs to be analyzed or processed.

stream(url, options)

parameter	Type	Description
ws_url	string	The WebSocket URL provided by the ngrok TCP service.
options	object	An object containing optional parameters (listen_mode, voice_quality and stream_on_answer)
listen_mode	string	Specifies who hears the streamed audio. Options are callee, caller, or both
voice_quality	string	he desired voice quality in bits per second. Options are 8000, 16000
listen_mode	boolean	Whether to start streaming after the call is answered.The default value is false.

Example Implementation

• Here’s how to implement the AI streaming feature to send real-time audio from both participants (caller and callee) to a WebSocket server:

Node.js

const { PiopiyAction } = require("piopiy");

const action = new PiopiyAction();

const ws_url = "ws://example.tcp.ngrok.url";

const options = {
  listen_mode: "caller",
  voice_quality: "8000",
  stream_on_answer: true,
};

action.stream(ws_url, options);

Python

from piopiy import Action
# Create an instance of the Action class
action = Action()

ws_url = "ws://example.tcp.ngrok.url"

options = { "listen_mode": "caller", "voice_quality": "8000", "stream_on_answer": True }

action.stream(ws_url, options)