Voice.ai Voices#

✨ Features#

• 9 Voice Personas - Carefully curated voices for different use cases
• 11 Languages - Multi-language synthesis with multilingual model
• Streaming Mode - Real-time audio output as it generates
• Voice Design - Customize with temperature and top_p parameters
• OpenClaw Integration - Works with OpenClaw's built-in TTS

---

⚙️ Configuration#

Set your API key as an environment variable:

bash

export VOICE_AI_API_KEY="your-api-key"

Get your API key: Voice.ai Dashboard

---

📦 Installation#

No install step is required. This skill bundles a Node.js CLI and SDK (no external npm dependencies).

🧩 Key Files#

• scripts/tts.js - CLI entrypoint
• voice-ai-tts-sdk.js - Node.js SDK used by the CLI
• voices.json - Voice definitions used by the CLI
• voice-ai-tts.yaml - API specification
• package.json - Skill metadata for tooling

Security Notes#

See SECURITY.md for the full security and privacy overview.

This skill:

• Makes outbound HTTPS requests only to https://dev.voice.ai
• Reads local files: voices.json
• Writes audio output to the --output path (default output.mp3)
• Does not execute shell commands and does not modify system configuration files

🌐 API Endpoint#

The SDK and spec use https://dev.voice.ai, which is the official Voice.ai production API domain.

---

🤖 OpenClaw Integration#

OpenClaw can invoke the CLI script directly if your environment exposes VOICE_AI_API_KEY. Use the /tts commands as configured by your OpenClaw installation.

---

📝 Triggers#

These chat commands work with OpenClaw:

CommandDescription/tts <text>Generate speech with default voice/tts --voice ellie <text>Generate speech with specific voice/tts --stream <text>Generate with streaming mode/voicesList available voices

Examples:

text

/tts Hello, welcome to Voice.ai!
/tts --voice oliver Good morning, everyone.
/tts --voice lilith --stream This is a long story that will stream as it generates...

---

🎙️ Available Voices#

VoiceIDGenderPersonaBest Forellied1bf0f33-8e0e-4fbf-acf8-45c3c6262513femaleyouthfulVlogs, social contentoliverf9e6a5eb-a7fd-4525-9e92-75125249c933malebritishNarration, tutorialslilith4388040c-8812-42f4-a264-f457a6b2b5b9femalesoftASMR, calm contentsmoothdbb271df-db25-4225-abb0-5200ba1426bcmaledeepDocumentaries, audiobooksshadow72d2a864-b236-402e-a166-a838ccc2c273maledistinctiveGaming, entertainmentsakura559d3b72-3e79-4f11-9b62-9ec702a6c057femaleanimeCharacter voiceszenithed751d4d-e633-4bb0-8f5e-b5c8ddb04402maledeepGaming, dramatic contentfloraa931a6af-fb01-42f0-a8c0-bd14bc302bb1femalecheerfulKids content, upbeatcommanderbd35e4e6-6283-46b9-86b6-7cfa3dd409b9maleheroicGaming, action content

---

🌍 Supported Languages#

CodeLanguageenEnglishesSpanishfrFrenchdeGermanitItalianptPortugueseplPolishruRussiannlDutchsvSwedishcaCatalan

Use the multilingual model for non-English languages:

javascript

const audio = await client.generateSpeech({
  text: 'Bonjour le monde!',
  voice_id: 'ellie-voice-id',
  model: 'voiceai-tts-multilingual-v1-latest',
  language: 'fr'
});

---

🎨 Voice Design#

Customize voice output with these parameters:

ParameterRangeDefaultDescriptiontemperature0-21.0Higher = more expressive, lower = more consistenttop_p0-10.8Controls randomness in speech generation

Example:

javascript

const audio = await client.generateSpeech({
  text: 'This will sound very expressive!',
  voice_id: 'ellie-voice-id',
  temperature: 1.8,
  top_p: 0.9
});

---

📡 Streaming Mode#

Generate audio with real-time streaming (recommended for long texts):

bash

# Stream audio as it generates
node scripts/tts.js --text "This is a long story..." --voice ellie --stream

# Streaming with custom output
node scripts/tts.js --text "Chapter one..." --voice oliver --stream --output chapter1.mp3

SDK streaming:

javascript

const stream = await client.streamSpeech({
  text: 'Long text here...',
  voice_id: 'ellie-voice-id'
});

// Pipe to file
stream.pipe(fs.createWriteStream('output.mp3'));

// Or handle chunks
stream.on('data', chunk => {
  // Process audio chunk
});

---

🔊 Audio Formats#

FormatDescriptionUse Casemp3Standard MP3 (32kHz)General usewavUncompressed WAVHigh qualitypcmRaw PCM audioProcessingopus_48000_128Opus 128kbpsStreamingmp3_44100_192High-quality MP3Professional

See voice-ai-tts-sdk.js for all format options.

---

💻 CLI Usage#

bash

# Set API key
export VOICE_AI_API_KEY="your-key-here"

# Generate speech
node scripts/tts.js --text "Hello world!" --voice ellie

# Choose different voice
node scripts/tts.js --text "Good morning!" --voice oliver --output morning.mp3

# Use streaming for long texts
node scripts/tts.js --text "Once upon a time..." --voice lilith --stream

# Show help
node scripts/tts.js --help

---

📁 Files#

text

voice-ai-tts/
├── SKILL.md              # This documentation
├── README.md             # Quick start
├── CHANGELOG.md          # Version history
├── LICENSE.md            # MIT license
├── SECURITY.md           # Security & privacy notes
├── voices.json           # Voice definitions
├── voice-ai-tts.yaml     # OpenAPI specification
├── voice-ai-tts-sdk.js   # JavaScript/Node.js SDK
├── package.json          # OpenClaw metadata
├── scripts/
│   └── tts.js            # CLI tool

---

💰 Cost & Usage#

Voice.ai uses a credit-based system. Check your usage:

javascript

// The SDK tracks usage via API responses
const voices = await client.listVoices();
// Check response headers for rate limit info

Tips to reduce costs:

• Use streaming for long texts (more efficient)
• Cache generated audio when possible
• Use appropriate audio quality for your use case

---

🔗 Links#

• Get API Key - Sign up and get your key
• API Documentation - Full API reference
• Voice Library - Browse all voices
• API Reference - Endpoint details
• Pricing - Plans and credits

---

📋 Changelog#

v1.1.4 (2026-02-16)#

• Declare VOICE_AI_API_KEY as primary env var in metadata

v1.1.3 (2026-02-16)#

• Remove voice-sample upload features from the published bundle to reduce privacy risk
• Require VOICE_AI_API_KEY via environment variable only

v1.1.2 (2026-02-16)#

• Added SECURITY.md and LICENSE.md for provenance and transparency
• Restricted SDK transport to HTTPS only

v1.1.1 (2026-02-16)#

• Packaging metadata improvements for ClawHub import (bin/files metadata)

v1.1.0 (2026-02-16)#

• Declared required credentials in metadata
• Documented the production API endpoint domain
• Renamed voice personas for IP-safe labeling
• Added voices.json for voice data

v1.0.0 (2025-01-31)#

• Initial release
• 9 curated voice personas
• 11 language support
• Streaming mode
• Voice design parameters
• Full SDK with error handling
• CLI tool

---

🛠️ SDK Quick Reference#

javascript

const VoiceAI = require('./voice-ai-tts-sdk');
const client = new VoiceAI(process.env.VOICE_AI_API_KEY);

// List voices
const voices = await client.listVoices({ limit: 10 });

// Get voice details
const voice = await client.getVoice('voice-id');

// Generate speech
const audio = await client.generateSpeech({
  text: 'Hello, world!',
  voice_id: 'voice-id',
  audio_format: 'mp3'
});

// Generate to file
await client.generateSpeechToFile(
  { text: 'Hello!', voice_id: 'voice-id' },
  'output.mp3'
);

// Stream speech
const stream = await client.streamSpeech({
  text: 'Long text...',
  voice_id: 'voice-id'
});

// Delete voice
await client.deleteVoice('voice-id');

---

❓ Troubleshooting#

ErrorCauseSolutionAuthenticationErrorInvalid API keyCheck your VOICE_AI_API_KEY``PaymentRequiredErrorOut of creditsAdd credits at voice.ai/dashboardRateLimitErrorToo many requestsWait and retry, or upgrade planValidationErrorInvalid parametersCheck text length and voice_id

---

Made with ❤️ by Nick Gill