Kakaia strives to be a personal assistant supporting voice commands without sending any data to third-parties.
This client is still just a proof of concept. It currently makes hard-coded assumptions based on my personal development environments.
The client requires that you are also running the Kakaia engine on your local network. It provides a simple UI for recording your voice from an Apple Watch, sending the audio file to the Kakaia engine to convert to text.
The app is currently just a proof of concept, and has 4 simple screens:
The app is not recording audio, but is standing by to record audio as soon as you press the big green button.
The app is recording audio:
Currently the app records continuously until you press the big red button.
The app is encoding the audio, sending it to the Kakaia engine, and waiting for a response:
The Kakaia engine recognizes a few spoken commands. Here's the output from it recognizing a simple math request.
Once the Kakaia engine returns the audio as text, the Kakaia text at the top of the screen turns into a button. Click it to see the text version of your spoken audio:
The app is currently a proof of concept with limited functionality. I greatly appreciate contributions!
Kakaia watchOS app:
- provide a simplistic UI for recording voice (done)
- send audio recording to API (done)
- wait for conversion (done)
- display text version on app screen (done)
- add configuration (currently it's hard-coded to my environment)
Add support for setting timers.
Kakaia watchOS app:
- parse returned JSON, set timer when the command is received (done)
- display errors if no command was matched
- add complication for recording audio from watch-face
- submit to App Store
See Kakaia engine Readme.
Contributions welcome!