What's new in Kiosk Satellite 2026.9.98
This release lets announcements and alarms speak in the language and voice you pick, and makes the announcement chime as loud as the words that follow it. It changes the ESPHome announce action, so automations and scripts that call it need two new fields.
Breaking Change: ESPHome Announce Action
- The announce action has two new fields. The ESPHome announce action now takes
tts_languageandtts_voice. Home Assistant requires every field of an ESPHome action, so existing automations and scripts that call it fail until both fields are added. - Update your announcement automations. Add
tts_language: ""andtts_voice: ""to keep speaking with the language and voice set on the kiosk. Fill them in to choose a language and voice for one announcement without changing the kiosk's settings. - Use the IDs the pickers show. Both fields take the value exactly as the engine lists it, not the display name. For Piper that is
es_MXfor the language andes_MX-claude-highfor the voice, notes-mxorclaude. The Language and Voice pickers show each ID under its name. When Home Assistant cannot speak with the language or voice, the kiosk speaks the message with the engine's defaults instead.
New Feature: Text to Speech Language and Voice
- Pick the language and voice. Announcements under ESPHome and Settings, Alarms have a new Text to Speech group with the engine, a Language picker and a Voice picker, on the device and in the remote admin. Both list what Home Assistant has for the picked engine. Voices follow the picked language, or the Home Assistant language when Language is Default. Default leaves the choice to the engine, and engines with no voices to pick, like Google Translate, offer only Default.
- Alarms speak with your voice. An alarm set to speak says its phrase with the Text to Speech language and voice from Settings, Alarms. Changing the voice makes a new recording instead of replaying the old one.
- Picking another engine starts fresh. A new engine resets Language and Voice to Default, since another engine's languages and voices do not carry over.
Improvements
- The announcement chime is as loud as the TTS. The chime played at the notification volume on its own audio stream, while the words played at the media volume, often through the call audio route that some devices play much louder. The built-in chime was also made at about half the level of the speech. The chime now plays just before the words on the same audio track at the same volume, and the built-in chime is as loud as text to speech audio. A custom chime from the sounds folder plays the same way.
- The remote admin stays put while you pick. Picking a text to speech engine, language or voice in the remote admin updates those rows in place instead of redrawing every settings page.