Lokutor JavaScript SDK

The official JavaScript/TypeScript SDK for interacting with the Lokutor API. Optimized for ultra-low latency text-to-speech and real-time voice chat.

Features

• Ultra-Low Latency: Built-in jitter buffering and scheduling for smooth audio playback.
• Voice Chat: High-level API for bidirectional voice communication with automatic interruption handling.
• Viseme Support: Synchronize 3D avatar animations with real-time viseme events.
• Cross-Environment: Works in modern Browsers and Node.js.
• Type Safe: First-class TypeScript support.

Installation

npm install @lokutortts/sdk

Quick Start

Initialize the Client

import { LokutorClient } from '@lokutortts/sdk';

const lokutor = new LokutorClient({
  apiKey: 'your_api_key'
});

Text-to-Speech (TTS)

// Standard Synthesis
const { audio, visemes } = await lokutor.tts.synthesize("Hello world", {
  voiceId: 'F1',
  quality: 'high'
});

// Stream Synthesis (NDJSON)
for await (const chunk of lokutor.tts.stream("Hello world", { includeVisemes: true })) {
  console.log("Received chunk with visemes:", chunk.visemes);
}

Voice Chat (Web/Browser Only)

const session = lokutor.voiceChat.createSession({
  voiceId: 'M1',
  quality: 'ultra_fast',
  includeVisemes: true
});

session.on('connected', () => console.log("Connected to Lokutor!"));
session.on('agentSpeaking', () => console.log("Agent started speaking"));
session.on('viseme', (v) => animateAvatar(v.id));
session.on('interrupted', () => stopAnimations());

await session.start();
await session.startTranscription(); // Starts mic and sending audio to Lokutor

Audio Utilities

The SDK includes LokutorPlayer and LokutorRecorder to handle the heavy lifting of Web Audio API.

import { LokutorPlayer, LokutorRecorder } from '@lokutor/sdk';

const player = new LokutorPlayer();
const recorder = new LokutorRecorder();

// Player handles PCM chunks and jitter buffering automatically
await player.playChunk(pcmData, 22050);

Documentation

• API Reference
• Voice Catalog