@dtelecom/server-sdk-node

Node.js RTC SDK for dTelecom — join rooms as a WebRTC participant for AI bots.

Features

• WebRTC participant — join dTelecom rooms as a bot
• Audio receive — subscribe to remote audio as PCM16 streams (16kHz for STT)
• Audio send — publish PCM16 audio as Opus (from TTS or any source)
• Data channels — send/receive messages via reliable or lossy channels
• Pure TypeScript WebRTC — uses werift (no native WebRTC deps)
• Opus codec — uses @discordjs/opus (native libopus bindings)

Install

npm install @dtelecom/server-sdk-node @discordjs/opus

Quick Start: Echo Bot

import { Room, AudioSource, AudioStream, AudioFrame, LocalAudioTrack } from '@dtelecom/server-sdk-node';

const room = new Room();
await room.connect('wss://my.dtelecom.org', token);

// Create audio output
const source = new AudioSource(16000, 1);
const track = LocalAudioTrack.createAudioTrack('echo', source);
await room.localParticipant.publishTrack(track);

// Echo received audio back
room.on('trackSubscribed', async (remoteTrack, pub, participant) => {
  const stream = remoteTrack.createStream(16000, 1);
  for await (const frame of stream) {
    await source.captureFrame(frame); // echo back
  }
});

Architecture

@dtelecom/server-sdk-node
├── SignalClient          WebSocket + protobuf to SFU
├── RTCEngine             2 PeerConnections (publisher + subscriber)
├── Room                  participant management, events
├── LocalParticipant      publish tracks + data
├── RemoteParticipant     subscribed tracks
├── AudioSource           PCM16 → Opus → RTP → publish
├── AudioStream           subscribe → RTP → Opus → PCM16
└── DataChannel           reliable + lossy messaging

API

Room

const room = new Room();
await room.connect(url, token, { autoSubscribe: true });

room.on('participantConnected', (participant) => { });
room.on('trackSubscribed', (track, publication, participant) => { });
room.on('dataReceived', (data, participant, kind, topic) => { });
room.on('disconnected', (reason) => { });

await room.disconnect();

Audio

// Publish audio
const source = new AudioSource(16000, 1);
const track = LocalAudioTrack.createAudioTrack('bot-audio', source);
await room.localParticipant.publishTrack(track);

const frame = new AudioFrame(pcm16Data, 16000, 1, samplesPerChannel);
await source.captureFrame(frame);

// Receive audio
room.on('trackSubscribed', async (remoteTrack) => {
  const stream = remoteTrack.createStream(16000, 1);
  for await (const frame of stream) {
    // frame.data = Int16Array (PCM16, 16kHz, mono)
  }
});

Data Channels

import { DataPacket_Kind } from '@dtelecom/server-sdk-node';

await room.localParticipant.publishData(
  new TextEncoder().encode('hello'),
  { kind: DataPacket_Kind.RELIABLE, topic: 'chat' }
);

room.on('dataReceived', (data, participant, kind, topic) => {
  console.log(new TextDecoder().decode(data));
});

Examples

# Echo bot — echoes received audio back
npx tsx examples/echo-bot.ts --room test

# Record audio — saves remote audio to PCM file
npx tsx examples/receive-audio.ts --room test --output recording.pcm

# Play audio — sends PCM file into room
npx tsx examples/send-audio.ts --room test --input hello.pcm

Dependencies

| Package | Purpose | Native? | |---|---|---| | werift | WebRTC (ICE, DTLS, SRTP) | No (pure TS) | | @discordjs/opus | Opus encode/decode | Yes (libopus) | | ws | WebSocket client | No | | protobufjs | Protobuf encode/decode | No |

License

Apache-2.0