Release Announcement
Changes
Updates to resolve a few bugs and improve compatibility with smaller installs.
Added
- Added nickname option to devices/containers
- Dependency on stt (a.k.a. “coqui” which is a replacement for deepspeech)
- Added new parameters for Speaker to be able to integrate with mimic3
Modified
- (CRITICAL) Fixed skill inclusion breaking runtime due to missing “create_skill()” attributes
- Cleaned up documentation on inclusion of libraries (added python3-venv and removed traceback)
- Corrected documentation on PyAudio library installation
- (CRITICAL) Fixed inclusion of missing files in PyPi build
- PyPi integrations updated and streamlined build
- Modified versioning storage/processing
- Updated to a basic README for PyPi download page
- Multiple bugs fixed in KasaPlug for local, direct plug access
- Bug fix for isAlive() to is_alive()
- Renamed all objects to support “kenzy” (and related variants)
- Updated “ask” function to start timeout after the originating utterence ends (rather than when it starts)
- Documentation for installing on Ubuntu 22.04 LTS (pyaudio and libfann source package installation workarounds)
- Download Models option now defaults to tflite format and pulls the Coqui base models from the Coqui Model Zoo
- Moved all request/response processing into Skills and removed hardcoded responses
- Removed hard dependency on padatious library
- Fixed bug in skills processing for multiple intents
- Updated device callbacks to use the GenericDevice naming convention
- Updated device settings to allow for store/update on the fly
- Updated container settings to allow for store/update on the fly
- Adjusted where version information is stored
Removed
- Dependecy on deepspeech
- Documentation dependency on padatious (libfann related issues for auto-build in readthedocs API)
Help & Support
Installation instructions and documentation is available at https://docs.kenzy.ai