He means that using a longer sample loop, the interpolation will be less noticeable and the output have more trebble relative to the note's tone. HOWEVER this also means your sample has a lower pitch at a given playback rate, and as such,that the highest note reachable (with frequency 0x3FFFF) is lower. In other word, you increase the sample rate, this boosts the treble but you cannot reach higher tones. This is holds true if any kind of interpolation is present, not only gaussian interpolation - so most synths will be affected by this.
To get a really crisp sound you'd need both : A sample who's sample rate is not lower than needed (so the pitch is at least 0x10000) and a trebble-boost when encoding.