A low pass filter @ 100- 150 hz or so will take out everything above the bass tracks, but you will also lose all the midrange and trebble parts. There is no way to lose just the vocals without losing the other midrange and trebble, without going back to the orginal studio multitracks that have the individual instuments/voices on different tracks. Human voice is mostly in the 300 hz - 3500 hz range.
Get the bass tracks with the low pass filter, get some of the full range music tracks from portions of the songs without vocals, copy, loop and make your own mix.
I too have noticed that some Timberlake songs have really good tracks, till that pvssy Timberlake comes in and ruins it. I give him credit for hiring good programers.