Thanks.
I will try to explain simply, but it is not easy. The interpolation filter is an FIR filter. This has delayed audio data. Each delayed data is multiplied by a coefficient, and all these multiplications are summed together, and the summed result is the interpolated value. Now the delayed data is held in a delay line, and is tapped into by a multiplier - hence the term taps. The coefficient is calculated by an algorithm, and in Hugo this is the WTA algorithm. This has been developed over many years, and what I am trying to do is maximize the sound quality performance, by improving the way the timing is reconstructed.
Now if you look at the mathematics, if you want to reconstruct the original bandwidth limited perfectly (that is all the timing of transients will be perfectly reproduced) then you need an infinite number of taps for the filter. When you start reducing the number of taps, then the timing accuracy degrades, which has very important subjective consequences. Hugo has the highest tap length filter of any DAC available at any price, which as you identify, is one substantial reason for it's naturalness. With Hugo, it has a 26,368 taps. This means for every sample of CD data, there are 26,368 multiplications and additions going on. To do this, I use 16 customized parallel DSP cores, each one running at 208 MHz. So there is a lot of DSP horsepower in Hugo.