Thanks for playing with the extension!
In the early development of the app I remember I got a lot of problems implementing the continues recognition feature. As you have supposed they come from Google Speech API.
- The speech recognition service stops when detecting silence for a few minutes
- The speech recognition service uses a secure real time connection that can be seriously affected by any network issue.
I remember have worked hard in order to get a stable service, here some solutions I have implemented:
- Automatically restart the service when stop working
- Detect the last time the user has talked, trigger an automatically restart after a few minutes
The google chrome speech recognition is free as Handsfree for Web so far I haven't hear about a pay service.
Things that might improve your experience,
- Try the "Press Ctrl before say every command" mode (it isn't 100% hands free)
- We are working in a third mode, where you can start the recognition with a command instead than pressing a key
In the future we might start using another Google speech recognition service, you can see some details here: https://cloud.google.com/speech-to-text/
The adoption of this service is not straight forward, is pay service and I'm not sure that will improve the speech recognition.
I'm not sure if I've answered the questions or concerns ?