@absolutejs/voice-elevenlabs

ElevenLabs text-to-speech adapter for @absolutejs/voice.

Install

bun add @absolutejs/voice @absolutejs/voice-elevenlabs

Setup

import { elevenlabs } from '@absolutejs/voice-elevenlabs';

const tts = elevenlabs({
	apiKey: process.env.ELEVENLABS_API_KEY!,
	voiceId: 'JBFqnCBsd6RMkjVDRZzb',
	modelId: 'eleven_flash_v2_5',
	outputFormat: 'pcm_16000'
});

The adapter uses ElevenLabs' official streaming text-to-speech endpoint and emits normalized audio events with PCM chunks.

If you need a warm persistent TTS session for lower turn startup latency, you can opt into ElevenLabs' WebSocket transport:

const tts = elevenlabs({
	apiKey: process.env.ELEVENLABS_API_KEY!,
	voiceId: 'JBFqnCBsd6RMkjVDRZzb',
	modelId: 'eleven_flash_v2_5',
	outputFormat: 'ulaw_8000',
	transport: 'websocket',
	websocket: {
		autoMode: true,
		inactivityTimeoutSec: 180
	}
});

What It Maps

• HTTP chunked audio stream -> audio
• warm WebSocket stream-input session -> audio
• request failures -> normalized error
• adapter close -> normalized close

This adapter currently supports PCM output formats for compatibility with @absolutejs/voice audio events.

Options

Supported options include:

• apiKey
• voiceId
• modelId
• outputFormat
• languageCode
• voiceSettings
• seed
• enableLogging
• optimizeStreamingLatency
• transport
• websocket

API Key

Set ELEVENLABS_API_KEY in your runtime environment, or pass the key explicitly in the adapter config.

ElevenLabs references used for this adapter design:

• https://elevenlabs.io/docs/api-reference/text-to-speech
• https://elevenlabs.io/docs/api-reference/streaming
• https://elevenlabs.io/docs/capabilities/text-to-speech