Selection of with voice input

Question

A raised question from How can I use selection-of with voice input in a Bixby input-view?. As the answer from this question,

Currently, there is no keyword support to map "home" as "first one", if that is what you are asking. However, if you model your JS file to take "home" as a valid input argument, and map it to the correct email address, the user can just say "home" during input selection.

For now, is there any way yet to map option for selection-of to input like "first one" that is already been build? And for modelling option in JS file, is there any references for it?

[Edit] I find out it can be achieved by adding appropriate training for the concept you related to the option for selection-of, not sure if this is the way. What if user say out word that's similar to the word, where should i add in more option for it? Is it inside a vocabulary file or add unlimited possibilities into js file like this And i think this is kind of related to the discussion in this topic 'Illegal Plan' message when tagging training data with Vocab types

Or should i use enum for each selection as stated in post How do I match a enum symbol in a result view? or Creating selection-of from enum?

Answer 1

Yes, basically by adding trainings to prompt, and using enum with vocab, often quiz capsules can get voice input of A, B, C, D mapped correctly, including "Ace, Bee, See" (wrong ASR result from voice).

I would recommend checkout the QuizIt capsule in our marketplace. It is a good example of mapping voice input to the selection-of in input-view.

One downside is the amount of training, developers need to training all the possible input. One implementation is to standardize choice. Consider a board game type of capsule where player are facing choices such as, "You are now standing in front of a door. What would you want to do?" One possible implementation is to do something like this: "Option one: Break open it.", "Option two: turn away", "Option three: try the key you just found" (option three will only present itself if you got a key). Therefore only training and matching "Option X".