Rock Labs: Voice Commands Published May 2, 2016 This post is first in what we hope will be a series of 'lab reports' of new technologies we're playing with. The goal of these reports is to share what we're working on with the goal of getting early feedback and ideas on implementation. Letʼs jump into our first one. The HTML5 Speech Recognition API allows JavaScript to have access to a browserʼs audio stream and convert it to text. In this sample we've used that capability to issue voice commands to Rock. Take a look. As stated this is all done without plugins, browser extensions, Java. and of course Flash (really you thought we'd go there...). Currently, Chrome is the only browser to implement it, but that should change in the future. Thoughts on usage in Rock: Page Navigation : We could possibly add a 'voice hint' field to each page to respond to ʼshow (me) ______'. Also could add a separate field for 'navigate command' for supporting things like 'add a new family' or ʼsend a new communication'. Dictation : As the Google blog below shows, this can be very handy for voice dictation. Filtering / In Page Nav : For certain 'analytics' pages, it could be powerful to add voice filtering for things like 'filter by ____ campus'. That said, there are some issues. Below is a short summary of some of the challenges: Itʼs good, but not perfect. You sometimes need to repeat yourself, which is somewhat awkward. Limited browser support. Doesn't work well when you'r're talking then giving it a command. It doesn't have a way to break up the conversation. Perhaps adding an 'OK Rock' separator would help? Doesn't work well in a crowded room with other vocies. Not a huge one, but there are some limitations if you are not running SSL. The conversation on this is just starting. Head over to our Slack community and discuss your thoughts in the new 'labs' channel. Additional Resources: The HTML5 Speech Recognition API Google Developer Blog annyang!