Visemes Are Not Even Close

Version 4.4

Sometimes I import my dialog.  Sometimes I record directly.  Either way, the visemes that appear are typically not even close.  Example: What I said "This is a" Visemes that appeared W F S W F. 

I just recorded each letter.  The mmm gave me S E W D.  Turned the number of visemes down in preferences and tried again.  The mmm gave me Woo.  The uh gave me Woo L.  The D gave me M Woo.  S and F worked.  Oh gave me L Woo.  I don't know what's going on.  The playback of my voice sounds clear and I enunciated well.

I just know that, with the amount of lip sync I do, the time involved in correcting every sound I make is too much.