I have had a number of posts asking me about DSD, how it is done within Hugo etc. A constant idea is that DSD should remain native and hence "pure" and that this would be best. Now I sympathise with this idea, as if it were analogue then one would want to keep everything as simple and direct as possible, as each analogue component you add, you degrade transparency. But this rule simply does not apply in the digital domain, as it is possible to have processing that is perfectly transparent (just to stop there - although a digital module can be made perfectly transparent, it is categorically not easy to do so).
So I need to explain why native DSD is a bad idea inside the DAC. Firstly, Hugo needs to see the original file that you can get hold off, so if it was originally PCM, use that, if it was DSD, feed that to Hugo via DoP. So my comments about native DSD apply about inside the DAC, something people do not see.
Firstly, some history. I first started getting involved in designing DAC's in 1989, when I heard Phillips Bitstream DAC the SAA7320. Compared to multi-bit DAC's at the time, it was a revelation - digital was starting to sound smooth and refined. Now these DACs were PDM types - that is 1 bit with 256 times oversampling - technically exactly the same as DSD but running at 256 times not 64. Now I started with these DAC's, made improvements, and I realised that the noise shapers were limiting resolution, so I started using multiple chips each with their own dither, to improve resolution. Noise shapers convert PCM to lower resolution data like 1 bit DSD. Also I found that the out of band noise from these noise shapers were overloading the analogue sections, giving noise floor modulation, making it sound harder. Also the DAC's were innately very sensitive to clock jitter. To try to solve these problems I designed the PDM1024, which had multiple noise shapers (improve resolution) and digital filtering (delay and add) to help with the jitter sensitivity and the out of band noise problems. Now the PDM1024 (early 90's now) gave a big step forward, but I could not resolve all of these problems. So I started developing Pulse Array, which was a multi-bit noise shape technology. To solve the noise shaper resolution problems, it runs at 2048 FS and is 5th order or better. This theoretically approaches 90dB more noise shaper resolution than PDM at 256 FS, and 150 dB more resolution than DSD 64. The Pulse Array modulation scheme also has the benefit in that it has much lower master clock jitter sensitivity than native DSD/PDM and, more importantly, has no jitter induced noise that is signal dependent as it is a constant clocking scheme - so it has no innate noise floor modulation. Also, by running at 2048 FS, the noise shaper noise at 1MHz is much lower - about 1000 times lower noise than usual DAC's. This means a simple analogue single stage with minimal filtering, so you get much more transparency. Also, the analogue active section has a much easier time, as RF induced noise floor modulation is fundamentally easier.
Now this happened in 1995. At the same time, silicon DAC designers were on a similar path - moving performance DAC's to multi-level noise shaping, away from single bit. At this time DSD started, which was moving in the opposite direction - instead of 256 FS it had reduced to 64 FS, simply because of data rate limitations on optical disks. Now as I have talked about in earlier posts, DSD has a major benefit - it does not have the big timing problems of PCM - but it suffers from much poorer resolution, and creates more distortion and noise than PCM. Using the WTA filter addresses (I won't say eliminates the timing issue because I think we need more taps than today to do that) the timing problems of PCM, giving you the potential of better resolution from PCM and overall better sound.
Now when DSD was first presented, it was claimed that processing could be maintained natively, that is if you wanted to add volume control or freq EQ, you could modify the bit-stream directly and re-noise shape the OP. But people quickly found out that this was not possible. When you re noise shape DSD 64 it very quickly degrades in terms of noise performance - you simply cannot connect 3 or 4 noise shaped stages together, unless you want awful performance. With regular PCM this is not the case, you can add as many stages together and it won't significantly change the measured performance. This is why DSD is initially recorded with PCM at 352.4 kHz - the DXD standard. Then it can be mixed, EQ etc, with minimal losses. Then finally it gets converted from DXD to DSD. And if you can get hold of DXD master recordings and the DSD you can hear the transparency losses of DSD (try 2L website they have free samples).
Now with Qute I did have a choice - I could easily use the DSD data and delay it, then feed it into the 4E DAC. But this would be a very bad thing to do, as DSD is -20dB at 100 kHz, and this noise is distorted, signal correlated, and would cause noise floor modulation in the OP stage. Also, it would be very master clock jitter sensitive, so you would get jitter induced noise floor modulation. The result would be back to the 90's, sub 100 dB dynamic range, distortion, noise floor modulation, whistles, pops and gurgle noises... together with poorer sound stage, poorer detail resolution and a hard aggressive sound quality.
So, in Qute it is digitally filtered, upsampled, filtered again. At this point you still have identical performance, as these steps can be done transparently, that is the audio spectrum is identical to the original signal. It is next fed into Qute's pulse array noise shaper, which will have a small price in transparency. But since this noise shaper runs at 2048 FS not 64 FS, at 20kHz, it is 10,000 times more capable of resolving signals than DSD. At 1kHz it is billions of times more capable than DSD. So the transparency loss is very very small compared with the enormous problems of using DSD natively. In Qute, I used a moving average filter, and the signal was always up-sampled, not decimated. At 100kHz the filter gave 50 dB worth of rejection.
Now in Hugo, we have a potentially much more serious problem with DSD, as Hugo has to do volume control and cross-feed EQ. This means it has to be converted to PCM, and at a rate the cross feed and volume control works at - which is 16FS. So the filtering was much more challenging now, as I had to decimate the signal too (make it a smaller sample rate). This meant a new design, as the Qute filter would have aliasing problems due to not enough stop band rejection - it needed much more than 50dB filtering that Qute had. So I decided on a sledge hammer approach to aliasing problems, by having 140 dB of rejection. This actually is much better than pro standard ADC aliasing filters, but that is another story. The other benefit of this filter was that it removed the DSD noise at 100 kHz, as it had 110 dB worth of rejection at that frequency. Now I could have the benefits of PCM with DSD in that out of band noise is non-existent.
So this filter was designed, verified, measured and listened too. Now I was worried about the listening tests, as I had not decimated DSD before. I listened to the Qute filter directly, and compared it to the Hugo decimating filter, with no volume or cross feed, with identical gain on the two filters. The Hugo DSD filter sounded very much better - DSD was a lot smoother and warmer and more natural, with no detectable loss in transparency or timing. So the benefits of the much better filtering at 100 kHz far outweighed the potential decimation errors by a very big factor.
So apologies if you could not keep up with the technicalities, it is very complex issue - but people kept asking. I hope that you have gotten a flavour of the difficulties involved. Simple and false statements like "native DSD is best" hide a very much more complex reality.