Powered by Google’s voice recognition.
Are there any open (source, data, model, what have you) voice recognition libraries out there? Good enough for this use case? All the major names in the field that spring to mind are proprietary.
I used Sphinx back in the day fairly successfully, it appears to still be active: