From what I found, -if you use soundfonts with software/sampler- you better check soundfonts property first. Most soundfont's samples saved in 16/44.1 format. If you set your sequencer to record (say) 24 bit 48KHz, then you'll notice minor sound differences. I can't say how it sounds, it just a bit different than it's original sound. To avoid this, record the soundfont track (or "bounce to track") in 16/44.1 first, then edit the result to 24 bit 48 KHz using external editor (Soundforge, Wavelab, etc.). In most of my case, it sounds better (closer to original) than purning sampler to record 24/48 out of 16/44.1 samples. What do you think?