Wasn't there an option to "train" Cortana to respond to your voice? If you "train" her to recognize your voice while you're yelling, then chances are you would get responses for a yell but not under normal speaking tones.
Also, ambient sound levels affect accuracy of recognition and response. Most phones come with a secondary microphone (usually located near the earpiece) that assists in noise-cancelling, but that requires the user to speak near the main mouth piece - otherwise the user's voice would be interpreted as "noise" since the volume in the main mouthpiece will be similar to the volume in the noise-cancelling secondary microphone, and the accuracy of recognition and frequency of response will decrease.
11-19-2015 12:09 PM