Hi pedal,
LMC has a ESS Sabre based DAC, and he has reported that he is very happy with the sound. I doubt he would say so if it made violins into flutes!
Not that you should take my expression on this with some grains of salt, but ...
But of course it is not so that any violin suddenly sounds like a flute. This is under certain conditions only, and you can well say that these conditions are the more difficult ones. I cannot explain these conditions, but can only refer to the ways a violin can be played, and hope you know what I mean. Just (try to) compare more or less resin on the stick (the horse hair), and the more resin is there, the more harmonics will be created. Why ? the stick (micro) stutters on the string. This might be a typicle example the violin may start to sound like a synth. When few resin is there, few stutters are there, and the viloin will sound more like a flute.
Neither is true in practice of course, but since a DAC may be profound in either way, the DAC may make it sound like that. The NOS DAC will impeed for the synth more, and the heavy OS DAC will impeed more for the flute.
All is exlained by my earlier pictures. The heacy OS DAC making a pure sine of pure squares, while the NOS DAC pertains the squares.
The latter is always the reality BUT, it now depends for 100% on how the DAC further deals with that;
I think I said it before, but the NOS DAC will repeat a square infinitly into the spectrum, meaning right over the Nyquist frequency, that causing aliasing NQ being the mirror, which is 22050 for 44100 sample frequency. So the squares bump back into the audible spectrum when nothing is done about it.
A lot more can be said about this (and some I talked about earlier in this topic), but the main point is and remains :
Harmonics are the results of square(ish) sounds, and they are as square as their origin, though lower in amplitude. When the original sound is turned into sines, the harmonics disappear, because a sine just does not have harmonics. Thus, a sound which is that sound because of harmonics, becomes a flute. A flute, because it just has no harmonics.
In between the above lines much more is to say and explain, but I think you get the grasp of it : a heavy oversampling DAC is just NOT ABLE to produce harmonics, if you only believe my earlier pictures. And they are not faked.
A violin is a "crackling" instrument, similar to a trumpet on that matter. That crackling is squares. Remove them, and both the trumpet and the violin start to sound like flutes.
I did not want my own ESS Sabre (Buffalo) to sound bad or anything, and the measuring I did afterwards. But it is just a fact that
a. inrtuments cannot be discerned for their origin, especially not in a larger orchestra;
b. how can they, if the harmonics creating them have been removed;
c. you need the reference to recognize it.
As I said earlier, maybe more in between the lines : nobody was complaining in advance of listening to the NOS1. I was there, they did not. That something was lacking (as described) is another matter (well, I think). But as soon as the reference was there, the ESS became, well, unlistenable. Not because I said so. Also, each and every time I noticed the nodding within two seconds after the start of a track, I mean it.
Lastly, to be serious on your remark an NOS DAC making a synth of a violin .. it should. It should just because of theories I briefly explained, and which IMO are not too difficult to understand. However, this is only about the part of aliasing (because of filterless) and that aliasing for 100% sure (because I know by meassuring) is only at very few places at an SPL that equals the listening level. So, this is 5 or 6 (by heard) sheer "points" in the audible band. Although this looks very wrong, in fact this is only about a few spikes like a speaker may (no, will !) let emerge here and there.
I put it my task to
a. measure properly;
c. interpret that towards how bad it is for actually hearing.
So far for me this is difficult, because I am not experienced on this, but on the other hand, actually nobody is, because official AES17 measurement measures the wrong things. Says who ? me. This is about THD+N of 0.0001% while a square became a sine.
I like to turn that into more truthful and meaningful figures. Just because it is as much audible as my theories tell.
Lastly, and I hope I am allowed to emphasize it this time, this time only music was compared which should just matter not all that much about OS/NOS. We did not compare trumpets, no cymbals, no sibilance reed instruments, no synthesizers, no high timbre male voices.
And *there* NOS really makes the difference.
Besides this, and this really matters too, we compared SPDIF only just to be honest in comparisons. I2S is just from another leage really ...
(but the ESS Sabre can theoretically do that too)
So far,
Peter