An Apple device running the Home Assistant app gains another job: processing speech for Home Assistant. A new voice tools server turns audio into text and text into spoken replies using Apple's on-device speech recognition and speech synthesis.
The service is off by default. Enabling it in Settings advertises the device to Home Assistant's Wyoming integration, the connection used for voice services. Processing stays on the Apple device, with transcripts and generated audio returned over the local network.
There is a practical limit: the server answers only while the app is open and stops when it goes into the background. Bruno Pantaleão Gonçalves contributes the feature, which includes settings for the listening port and lists of the recognition languages and voices available on that device.