I don't think that's fair to Ms Kelly, I don't really have a problem with her conversational style but I think however NPR processes her speech creates that problem. Or, probably more significantly, doesn't process her speech by not removing crackles and breathing. I'm curious how much real-time processing is done by other radio stations.

Another way to put it is that the more mainstream news personalities have developed their speech to be extremely artificial. NPR deliberately aims for more natural voices, and expect that listeners will make certain assumptions about personality from those voices.