Consider the phrases “recognize speech” and “wreck a nice beach.” If you say them out fast anyone might confuse the sentences. From this it is easy to see why NLU can experience difficulties when performing voice recognition. Name at least two issues a good voice recognition system must overcome to accurately identify natural human speech. Describe how these issues affect VR development.

Basically, they sound the same.

Explanation:

Because, when you have 2 sentences and they both sound the same the VR system recognizes only one thing because you're both saying them the same kind of way so it's really confusing.

Dialects- because native languages have many different dialects it would be hard to account for all of the varrences. Chinese is (one of) the densest languages in the world since there are literally hundreds of thousands of sub-varieties of the language. Trying to develop a software that can capture all of these varrences will be difficult because dialects change and mutate as people move in and out of regions.  

Loud environments- Since voice recognition systems are used in all sorts of environments it would be hard to account for the different background noise that will be heard and have to be filtered through. Creating a VR that can identify all noises will be virtually impossible.

Explanation: