Voices

Lumia supports four voice sources. They can be mixed freely; a follow alert uses Brian, a sub alert uses a marketplace voice, and a chatbot command uses a system voice.

System voices

The OS speech synthesizer: macOS has Siri voices, Windows has Narrator voices. System voices are always available with no network call. Quality varies by OS and installed language packs.

Voice id and label values come from your machine. Retrieve the full list at runtime via the Get Settings endpoint:

GET http://localhost:39231/api/settings?token=YOUR_TOKEN

The voice list is at data.options.tts.values.voices. Each entry has:

{
  "id": "Agnes",
  "label": "Agnes",
  "language": "en_US"
}

Pass the id or label as the voice field in any TTS call.

Lumia cloud voices

Lumia's hosted voice set, powered by Amazon Polly, gives you high-quality voices without any third-party account. Available to all Lumia users.

Well-known cloud voices:

Voice	Language
Brian	en-GB
Amy	en-GB
Emma	en-GB
Joanna	en-US
Kimberly	en-US
Salli	en-US
Justin	en-US
Joey	en-US
Nicole	en-AU
Russell	en-AU

These are referenced by name in the voice picker and in API calls. Pass the voice name exactly as shown.

Marketplace voices

The Lumia voice marketplace hosts community-uploaded voices. Browse and install them from the Lumia settings panel. Once installed, a marketplace voice appears in the voice picker alongside system and cloud voices.

Marketplace voices are referenced by their id from the settings voice list. Because they route through Lumia's hosted TTS proxy, they require an active internet connection.

Plugin voices (ElevenLabs, TTS Monster, and others)

Plugin voices bypass the standard TTS pipeline entirely. The plugin:

1. Intercepts the TTS request.
2. Sends the text to an external service (ElevenLabs API, TTS Monster API, etc.).
3. Receives audio bytes back.
4. Calls this.lumia.playAudio() to play them through Lumia.

Installing a plugin like ElevenLabs TTS or TTS Monster makes its voices available in the Lumia voice picker for every command and alert. No code change needed: configure the API key in the plugin settings, pick the voice.

See the example plugins:

• ElevenLabs TTS
• TTS Monster

Random voice

Pass "random" as the voice to select a random English voice from the available cloud pool on each firing:

{
  "type": "tts",
  "params": {
    "value": "Random voice every time!",
    "voice": "random"
  }
}

Checking available voices at runtime

GET http://localhost:39231/api/settings?token=YOUR_TOKEN

Parse data.options.tts.values.voices for the full list of { id, label, language } entries installed on the current machine.

To list only cloud voices, filter for entries where source === 'cloud'. Marketplace voices have source === 'marketplace'.