* Protobuf audio test
* Remove extraneous code
* Rework voice assistant pipeline
* Move variables
* Fix reading flags
* Dont directly put to queue
* Bump aioesphomeapi to 24.0.0
* Update tests
- Add more tests for API pipeline
- Convert some udp tests to use api pipeline
- Update fixtures for new device info flags
* Fix bad merge
---------
Co-authored-by: Michael Hansen <mike@rhasspy.org>
* Add ATTR_PREFERRED_FORMAT to TTS for auto-converting audio
* Move conversion into SpeechManager
* Handle None case for expected_extension
* Only use ATTR_AUDIO_OUTPUT
* Prefer MP3 in pipelines
* Automatically convert to mp3 on demand
* Add preferred audio format
* Break out preferred format
* Add ATTR_BLOCKING to allow async fetching
* Make a copy of supported options
* Fix MaryTTS tests
* Update ESPHome to use "wav" instead of "raw"
* Clean up tests, remove blocking
* Clean up rest of TTS tests
* Fix ESPHome tests
* More test coverage
* ESPHome Wake Word support
* Remove all vad code from esphome integration
* Catch exception when no wake word provider found
* Remove import
* Remove esphome vad tests
* Add tests
* More tests
* Change to "finished speaking detection"
* Add select entity to ESPHome for finished speaking detection
* Fix entity name
* Use vad select in stt stream
---------
Co-authored-by: J. Nick Koston <nick@koston.org>
* Add speech detection to esphome voice assistant
* Timeout after silence
Ensure events are sent before finish is called
* use va_version 3 instead of ESPHome version
* Convert repeated fixtures to factory
* Add some v3 tests
* Add conversation_id
* Bump aioesphomeapi to 13.8.0
* Fix missed buffering of detected chunk
* Alter log message
* Updates
* Spelling
* Fix return type
