Elevenlabs

ElevenLabs — Voice AI Platform

Generate speech, music, and sound effects via the ElevenLabs API. Supports 70+ languages, 22+ premium voices, and multiple models for different quality/latency tradeoffs.

Prerequisites

• ELEVENLABS_API_KEY environment variable must be set (get it from ElevenLabs dashboard → API Keys)
• For JavaScript: use @elevenlabs/elevenlabs-js package only
• For Python: use elevenlabs package

Models

Default recommendation: eleven_multilingual_v2 for production voiceovers, eleven_flash_v2_5 for real-time/interactive.

Quick Start

JavaScript/TypeScript

import { ElevenLabsClient } from "@elevenlabs/elevenlabs-js";
import { createWriteStream } from "fs";

const client = new ElevenLabsClient();
const audio = await client.textToSpeech.convert("JBFqnCBsd6RMkjVDRZzb", {
  text: "Hello, welcome to ElevenLabs!",
  modelId: "eleven_multilingual_v2",
});
audio.pipe(createWriteStream("output.mp3"));

Python

from elevenlabs import ElevenLabs

client = ElevenLabs()
audio = client.text_to_speech.convert(
    text="Hello, welcome to ElevenLabs!",
    voice_id="JBFqnCBsd6RMkjVDRZzb",
    model_id="eleven_multilingual_v2"
)
with open("output.mp3", "wb") as f:
    for chunk in audio:
        f.write(chunk)

cURL (batch scripting)

curl -X POST "https://api.elevenlabs.io/v1/text-to-speech/${VOICE_ID}" \
  -H "xi-api-key: $ELEVENLABS_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"text": "Hello!", "model_id": "eleven_multilingual_v2"}' \
  --output output.mp3

Popular Voice IDs

List all available voices:

voices = client.voices.get_all()
for voice in voices.voices:
    print(f"{voice.voice_id}: {voice.name}")

Voice Settings

Fine-tune voice output with these parameters:

from elevenlabs import VoiceSettings

audio = client.text_to_speech.convert(
    text="Customize my voice settings.",
    voice_id="JBFqnCBsd6RMkjVDRZzb",
    voice_settings=VoiceSettings(
        stability=0.5,
        similarity_boost=0.75,
        style=0.5,
        speed=1.0,
        use_speaker_boost=True
    )
)

See references/voice-settings.md for detailed guidance.

Language Enforcement

Force specific language pronunciation with ISO 639-1 codes:

audio = client.text_to_speech.convert(
    text="Bonjour, comment allez-vous?",
    voice_id="JBFqnCBsd6RMkjVDRZzb",
    model_id="eleven_multilingual_v2",
    language_code="fr"
)

Request Stitching (Long Audio)

When generating audio in multiple parts, use previous_text/next_text to avoid pops and tone shifts at boundaries:

audio1 = client.text_to_speech.convert(
    text="This is the first part.",
    voice_id="JBFqnCBsd6RMkjVDRZzb",
    next_text="And this continues the story."
)

audio2 = client.text_to_speech.convert(
    text="And this continues the story.",
    voice_id="JBFqnCBsd6RMkjVDRZzb",
    previous_text="This is the first part."
)

Output Formats

Streaming

For real-time applications:

audio_stream = client.text_to_speech.stream(
    text="This text will be streamed as audio.",
    voice_id="JBFqnCBsd6RMkjVDRZzb",
    model_id="eleven_flash_v2_5"
)
for chunk in audio_stream:
    play_audio(chunk)

See references/streaming.md for WebSocket streaming.

Batch Generation (Pipeline Pattern)

For generating multiple audio files from a script (voiceovers, courses, etc.):

import { ElevenLabsClient } from "@elevenlabs/elevenlabs-js";
import { writeFileSync } from "fs";

const client = new ElevenLabsClient();
const scenes = [
  { id: "intro", text: "Welcome to the show." },
  { id: "main", text: "Today we discuss AI." },
  { id: "outro", text: "Thanks for watching!" },
];

for (const scene of scenes) {
  const audio = await client.textToSpeech.convert("JBFqnCBsd6RMkjVDRZzb", {
    text: scene.text,
    modelId: "eleven_multilingual_v2",
  });
  // Collect chunks and write
  const chunks: Buffer[] = [];
  for await (const chunk of audio) chunks.push(Buffer.from(chunk));
  writeFileSync(`public/voiceover/${scene.id}.mp3`, Buffer.concat(chunks));
}

This pattern integrates directly with Remotion — see the remotion skill for the full ElevenLabs → Remotion pipeline.

Text Normalization

Controls how numbers, dates, abbreviations are spoken:

• "auto" (default): Model decides
• "on": Natural speech ("January fifteenth")
• "off": Literal ("zero one slash one five")

Tracking Costs

response = client.text_to_speech.convert.with_raw_response(
    text="Hello!", voice_id="JBFqnCBsd6RMkjVDRZzb", model_id="eleven_multilingual_v2"
)
print(f"Characters used: {response.headers.get('x-character-count')}")

Error Handling

Music Generation

Generate original music from text prompts (up to 10 min):

# Via inference.sh CLI
infsh app run infsh/elevenlabs-music --input '{
  "prompt": "Upbeat electronic music with synths",
  "duration_seconds": 30
}'

Sound Effects

Generate SFX from text descriptions:

# Via inference.sh CLI
infsh app run infsh/elevenlabs-sound-effects --input '{
  "prompt": "Dramatic thunder crack followed by rain",
  "duration_seconds": 5
}'

References

• Installation Guide — SDK setup for Python, JS, cURL
• Streaming Audio — WebSocket streaming patterns
• Voice Settings — Detailed parameter tuning guide