Filtering speech signal before framing

Hi everyone! I'm doing a classwork about Speech Recognition of isolated command and i'm doing the 1st step of PreProcessing. I've been able to record a word with mic and save data+Fs with wavwrite command. I've plotted the data and notice that there is a sinusoidal signal that overlap on the word signal (i think it's the 50Hz noise). How can i get rid of this noise? I've tried to use fir1 command specifing a window of [70 3200] to eliminate frequencies that can't be emitted by human voice but i don't know if i'm doing well. Then i was wondering to use a pre-emphasis filter with [1 -0.97]window parameter. I've read that the second parameter could be 0.95 too... how can i understand which is the right value? Can someone help me? Thank you so much!