SUSI.AI uses Raspi and the Respeaker Voice Hat. Successful user interactions depend a lot on the environment and skills being used. Changing the hotword sensitivity does not result in acceptable outcome as the device switches on very quickly even though the hotword was not very similar to the actually trained hotword. The goal of this project is to improve this feature.
Related Features
Improving hotword recognition
Implement automatic sound control
Automatically adjust sensitivity and volums
Required Knowledge
Linux OS, Package management, Voice Recognition, AI Algorithms, PocketSphinx, Flite
Why there are no ideas for Susi android. It had lots of severe bugs to solve and also some features like voice input, etc to be improved. Also, need to make UI a bit smoother and more interactive.
SUSI.AI uses Raspi and the Respeaker Voice Hat. Successful user interactions depend a lot on the environment and skills being used. Changing the hotword sensitivity does not result in acceptable outcome as the device switches on very quickly even though the hotword was not very similar to the actually trained hotword. The goal of this project is to improve this feature.
Related Features
Required Knowledge
Linux OS, Package management, Voice Recognition, AI Algorithms, PocketSphinx, Flite
Possible Mentors
Sanskar Jethi, Michael Christen, Hong Quan,
Repository
SUSI Linux: https://github.com/fossasia/susi_linux