0.8.0

0.8.0 (2026-09-21)

Feat

• speak live replies as auto, streaming or buffered (api_version 5)

0.7.0

0.7.0 (2026-09-19)

Feat

• name a delivery, and let one route carry every reply
• forward the words and let the app pace the reply

0.6.2

0.6.2 (2026-09-15)

Fix

• offer a regional voice to every pipeline of its language

0.6.1

0.6.1 (2026-09-15)

Fix

• send normalize_text only when the automation set it

0.6.0

0.6.0 (2026-09-15)

Feat

• send the language on every model, and leave the Chinese switches to the server

0.5.0

0.5.0 (2026-09-14)

Feat

• carry the language and the style instruction to the app

Fix

• config_flow: name an entry after its address, not after a version

0.4.0

0.4.0 (2026-09-13)

Feat

• sensor: report how many requests a reply took

0.3.1

0.3.1 (2026-09-13)

Feat

• tts: size a request in seconds, and cut a run-on where a reader would pause

Fix

• strings: name the grouped mode for what it does, and bound both claims to the engine

0.3.0

0.3.0 (2026-09-13)

BREAKING CHANGE

• cortex_tts.list_voices no longer accepts model. Pass the
  entity_id you intend to speak through, or no argument at all for every voice
  on every configured server.

Feat

• list_voices asks one entity, on that entity's own server

Fix

• tts: measure the margin before the late chunk lands, drop the gap sensor

0.2.1

0.2.1 (2026-09-13)

Fix

• config_flow: keep the example address out of the translated string