Voice Syntax

[DD1]

Reading the syntax guide I would be assured that the following Kinect voice phrase would require the word SET to be spoken in order for the command to work:

set [the] [Ground] [Office] Door Lock to Unlock

But if I use the Kinect attention phrase followed by:

Door Lock to Unlock

The command is initiated... is there a reason the word "SET" does not have to be used? If so, then why would it not be enclosed in brackets?

Thanks

[spud]

Reading the syntax guide I would be assured that the following Kinect voice phrase would require the word SET to be spoken in order for the command to work:

set [the] [Ground] [Office] Door Lock to Unlock

But if I use the Kinect attention phrase followed by:

Door Lock to Unlock

The command is initiated... is there a reason the word "SET" does not have to be used? If so, then why would it not be enclosed in brackets?

Thanks

if you set the log level to Trace, do you see the "Set" in the recognized phrase?
If not then it is a bug, but I doubt it.
If you feel that the Kinect plugin recognize some words that you don't say then increase the confidence threshold.

[DD1]

if you set the log level to Trace, do you see the "Set" in the recognized phrase?
If not then it is a bug, but I doubt it.
If you feel that the Kinect plugin recognize some words that you don't say then increase the confidence threshold.

Confidence threshold has nothing to do with the recognition of the phrase with the omitted word "SET"

It appears that the phrase recognition engine does not look at a strict structure of the captured voice command. In essence, if the phrase to be recognized is composed of 5 essential words in a Homeseer text box, the recognition engine hypothesizes (that is the term used in the log) that a captured voice command with 4 of those 5 words is "close enough". I can increase the threshold to .95 and still omit the word "SET". If I increase the threshold to 1.0, nothing works as the confidence level is never higher than .97 no matter how I speak the phrase.

As far as answering your question about seeing the word "SET" in the log, yes, the word "SET" is in the log even when I don't speak it. The log records the phrase as it was entered in the Homeseer text box and not what was actually spoken. The log never shows the words "EXACTLY" as I have spoken them. In fact, it seems to find words from phrases I have spoken an hour ago and parses them in at random when I speak phrases that omit more than a couple of words from an acceptable phrase.

Bottom line... the speech recognition engine can recognize a phrase even if a word (maybe multiple words?) are absent. Therefore, relying on strict phrase composition to run an event or control a device is a gamble.

Nice learning experience but it only reinforces my understanding of "nothing is 100% black and white." Nothing against your plugin... you are developing with the tools that Microsoft have given you.

Whether a light turns on with "Set the light to on" or "Light to on" is irrelevant if strict adherence to structure isn't a necessity. In fact, some people will be disappointed to learn that they have been saying an extra word or two thinking that they needed to match their phrase exactly as written.

That's how the plugin works for me... DISCLAIMER: Individual results may vary!!!