Text to speech
Settings → Text to speech reads messages aloud. Addons with permission can send text through the engine you configure here (for example donation alerts). You pick the engine, voices, and volume; API keys and local models stay in StreamKit+.
Common controls
| Setting | What it does |
|---|---|
| Enable TTS | Master switch |
| Speech engine | Piper (local), ElevenLabs (cloud), or Windows TTS |
| Volume | 0–200% (values above 100% are amplified in the main window) |
| Language, test phrase, Test voice | Check the current voice before going live |
If the addon does not specify a language, StreamKit+ detects English, Russian, or Ukrainian automatically (with a fallback to English).
Piper (default)
Local Piper voices. No account required.
- Download and manage ONNX voice models per language (English, Russian, Ukrainian).
- Optional CUDA acceleration on NVIDIA GPUs (Windows and Linux).
- Launch mode: start the worker immediately, or on first speak (saves RAM; first line may be slower).
- Speed and expressiveness sliders.
- A status indicator shows whether the worker is running.
If the Piper worker is missing from the install, the section explains that it could not start.
ElevenLabs
Cloud voices via the ElevenLabs API (eleven_multilingual_v2).
- Add one or more API tokens with character quotas. StreamKit+ picks the token with the most remaining quota (cached for a few minutes).
- Shared voice settings: stability, similarity, style, speed.
- Refresh limits from the panel when you need an up-to-date quota.
Needs a network connection and a valid token.
Windows TTS
System voices (SAPI). Windows only — the panel is disabled on Linux and macOS.
Configure a voice, rate, and volume per language.