I always dislike, 'which is best', type audio questions. What's best will vary by person. It will also vary by system context, as we don't listen to DAC boxes by themselves. This type of question also assumes that everyone responding has heard every possible candidate product vying for the title of best, which, of course, will not be true. That said, I would like to offer the following observation or two about the direction of leading edge DAC development. Should you have the patience, I have some rather rambling thoughts I'd like to share
It seems that vendors of the highest end products take one of two routes. They spend the great majority of their product's cost budget either on the digital/mixed signal processing side, or on the post conversion analog processing side. On the digital/mixed signal processing side, great effort and expense is typically applied to things such as increasing the dynamic range of 24-bit playback by a few extra dB, even though it is likely over 100dB to start with. Another common area of focus is trying to lower clock jitter from picosecond to femtosecond levels. Yet another area of focus is on delivering ever higher up/oversampling rates.
On the analog processing side, we see much the same efforts as are seen in analog preamps. Analog side implementation can become much more costly than that of the digital side, especially when transformer coulled tube circuits are employed, ala' Audio Note and Ypsilon. Both of those ultra expensive DACs feature digital processing which could fairly be described as being quite primitive. Alternatively, some of the top DACs that focus on the digital processing side feature IC op-amp based analog stages, such as those from Berkeley Audio Design and Bricasti. It would seem that more than one road can be taken to Rome.
So, why did I take the effort to lay out these rather well known facts? It is because I have a larger audiophile's concern (aside from stratospheric product pricing) that I feel applies to audio electronics seeking 'very best' status in general. Which is, that we seem to have arrived a the point of greatly diminishing returns (some would say we arrived there quite some time ago) with each subsequent increase in some common performance specification, or exotic part upgrade. The DAC situation mirrors what happens with amplification chain products, with their ever lower measured THD, noise and ever wider signal bandwidths, but things are even more specification focused with DACs today. However, where is the conmensurate audible benefit of this effort and expense? Is there ANY audible benefit?
This state of affairs is not uncommon across many types of products. Performance and features are steadily improved, if only by minuscule amounts, along established dimensions. What isn't asked enough, it would seem, is whether those established paths will eventually lead to fully closing the gap between current product performance and the users desire for product satisfaction. Perhaps, the current paths establishd for audio design in general, and DAC design in particular, will deliver us the home musical experience satisfaction we all seek but just can't seem to get sufficiently satisfied.
Would products having absolutely zero distortion, noise and jitter, infinite dynamic range and perfectly flat bandwidth then sound like real instruments? Some products are extremely close to delivering perfect specifications, yet are we even close to enjoying fully satisfying sound from them? Is some key factor not being measured, or it simply a case of the home stereo reproduction paradigm itself being irretrievably flawed?