How To Configure and Use Text-to-Speech

SheevChat can read chat and action-generated text through the desktop app or an active Overlay Builder browser source.

Choose a Voice Engine

Open Settings → Audio and choose:

  • System voice for voices already installed in Windows.
  • Piper for a lightweight local AI voice.
  • Kokoro for a higher-quality local AI voice.

Install Piper or Kokoro first under Settings → General → Third-party Dependencies. Select a voice and use the available test before relying on it live.

Configure Chat TTS

In Audio settings:

  1. Enable TTS.
  2. Choose the platforms and viewers that may be read.
  3. Set volume, voice, and any skip or queue behavior.
  4. Set the character limit from 50 to 2000.
  5. Save and test with a normal chat message.

Long text is divided into browser-safe speech segments so local system, Piper, and Kokoro voices can finish the complete message.

Use Settings → Filters → Filtered Accounts to exclude bots, relay accounts, or specific viewers from TTS.

Add TTS to an Action

Commands, Automations, Timers, and SheevPad can use Send to Overlay → Send TTS/Text Payload.

  1. Add the TTS action.
  2. Enter text or insert event variables.
  3. Select the global voice or an available action-specific override.
  4. Choose the intended overlay target when offered.
  5. Save and test the parent Command, Automation, Timer, or button.

Example:

{user} says {message}

Only use viewer-provided {message} text when the trigger and permissions make that safe for your stream.

Route TTS Through OBS or Streamlabs

Every active Overlay Builder source can receive SheevChat TTS and audio through the global overlay-audio path. A separate Audio Receiver widget is not required.

  1. Add the saved Overlay Builder URL as a Browser Source.
  2. Enable browser-source audio in OBS or Streamlabs.
  3. In Overlay Builder, select the overlay and use Test TTS.
  4. Trigger the real action and set its level in your broadcast-software mixer.

Legacy Overlay Audio Receiver widgets remain compatible, but do not add one to a new layout unless you specifically need its older channel-routing behavior.

Control Long or Repeated Speech

  • Keep the character limit appropriate for the trigger.
  • Add Command or Automation cooldowns to prevent spam.
  • Use permissions for viewer-submitted TTS.
  • Use TTS skip controls or configured hotkeys when speech must stop immediately.
  • Remember that queued items normally play sequentially rather than overlapping.

Common Problems

  • Piper or Kokoro is unavailable: Install its dependency, then return to Audio settings and choose a voice.
  • Overlay TTS is silent: Keep the Browser Source active, enable its audio, and run Overlay Builder's Test TTS.
  • Speech comes from the wrong place: Verify whether the action uses desktop playback or Send to Overlay.
  • A bot is read aloud: Add it to Filtered Accounts with the TTS scope enabled.
  • Messages are cut off: Increase the character limit and retest the selected voice engine.

See How To Manage Third-Party Dependencies for local voice installation and How To Set Up Overlay Audio for browser-source routing.