Here lists some issues when I was exploring Human-bot interaction.
|User makes longer pauses while speaking.|
Often with some sounds like “erm, uhm, eh” etc.
“Tom eats …(erm)... an apple”
|The system would stop after first words and consider this as the answer. For humans, if they only can rely on voice (let's say on the telephone) there would be probably following scenarios:|
1. He hears something he recognizes as fillers ('uhm', 'erm') and just waits
2. He knows that the sentence is not yet done and waits (because what user said is just the beginning of what he expects)
3. He interrupts the user and says something
|The user says something wrong and starts over|
“Tom eats the ... tom eats an apple”
|They user says acorrect answer and immediately repeats it again (maybe for his ownconfirmation, or because he thinks he wasn’t speaking clearly enough etc.)|
“Tom eats an apple ... tom eats an apple”
|The user repeats some words while saying the sentence, because he is simultaneously in theprocess of thinking|
“Tom eats.. eats.. eats.. an apple”