One thing I have found is that if a make a very non-fatiguing SD dac, then great recordings sound great, and poor recordings sound exactly like what they are, poor. However, if I dirty up the dac a little (say, an electrolytic somewhere instead of a good film cap) then the poor recordings are partially masked and they sound better, but the great recordings sound worse. So, IME it also depends on what recordings one likes to listen to.
The discretely (stair-stepped) analog output of a DAC chip contains both the desired original signal band, plus a series of repeating image-bands. Each image-band contains the same imformation as does the desired signal band, just progressively stepped higher in spectrum. More precisely, the image-bands manifest in mirror-imaged pairs, which are centered directly above and below multiples of the output sample rate. For example, with a 44.1KHz NOS output sample rate, the lower half of the first image-band pair extends from 44.1KHz down to 22.05KHz, while the upper half of that pair extends from 44.1KHz up to 64.15KHz. This mirrored-pair pattern repeats at the next multiple of the output sample rate, so, on each side of 88.2KHz, then again at 132.3KHz, and so forth.
This also means the OS is theoretically closer to the original signal than NOS. Strangely, I still prefer NOSCould it be because of the enriched harmonic content that is somehow pleasing to the ear?
Yes, OS is theretically closer to the original signal because it filters away the image-bands. That's what OS does, it's a digital filtering process. Remember, the image-bands are what produce the stair-stepped looking raw DAC output. Filter them away and the signal becomes smooth again. It is, by definition, reconstructed. I talked a little about this early in the thread in a post about how to think about interpolation.
My suspician is that you are correct. Let's hope that we are successful in conclusively establishing whatever is the best sounding technique.
I understand from this that you never heard a dac that it as transparent as what i am talking about.
No masking here, if the recording is bad, yes, you hear it but you hear also why... and then you can choose to enjoy the music badly recorded or not.
I can hear exactly how a mix is done and if the instruments are recorded seperately or not, multimike recordings and what effects have been used on each instrument. If this is bothering you (not meant personally here) then masking could be a choice but i see it as the "artists choice" (producers choice).
Until listening tests are performed, sharing such experience doesn't provide useful findings to people who weren't there.
There are recordings that are just bad, say, recorded to used tape, and where instruments are fighting each other in the same frequency space, and the mix has various other problems. The same recordings sound bad on vinyl too. To me some of those recordings can be more listenable if the system slightly remasters them. I'm not necessarily a purist in the sense of feeling recordings are sacred even if they are poorly done.
That's why subjective listening impression of one forum member is useless to another forum member.
Right. The job of hi-fi audio replaying electronics is to be transparent (fidelity) at high level, thus the term hi-fi. Since this is audio, what's audible is at stake. If this is marketing, then what's marketable (measurably lower distortion than what's already below hearing threshold) would be at stake.
In case of a really transparent dac you do not hear hear "the fighting" of the harmonics anymore but hear them as different instruments intertwine with each other...
I think i have meantioned something earlier about not beeing able to explain until heard first hand...
So who is to disagree on my findings unless having listened to this dac?
I also must trust what you have experienced to be true for you (again not meant personally but in general)...
Best wishes
I was speaking of a problem with musical arrangement, tracking, and mixing. Not the dac garbling up what is on the recording. I can hear the limitations of the ADCs used for digitizing recordings too. Some ADCs sound better than others; the sound of digital recordings is not just in the dac.
According to a friend who does high end audio design for a living, and has done so for the past 40 years, my dac design is in the same league as the big name high end dacs, but then again he said my dac is not at the top of that league. Still working on making it better
According to a friend who does high end audio design for a living, and has done so for the past 40 years, my dac design is in the same league as the big name high end dacs, but then again he said my dac is not at the top of that league. Still working on making it better
Last edited:
Mark, you make some important points, which I'd like to address.
It was obvious to me from the beginning that one reality we would have to accept is the lack of real scientific control over our listening experiments. This lack of control is necessitated for a number of reasons, such as, the fact that we are a group of internationally dispersed hobbyists, and that we are merely hobbyists of limited resources, and because the answers which we would seek inherently must be subjectively judged by listening.
While we don't have controls over everyone's system, it does help that we are not searching for system configuration and components which sound better than some other configuration. We are only attempting to find the technical reason(s) why NOS playback sounds different from OS playback. Not which sounds better, mind you. Of course, to directly participate in experiments requires that a person, at the least, possess both an OS and an NOS DAC. Better still, of course, would be a DAC wich has switchable OS and NOS modes, so that the entire system configuration remains identical in either mode. However, even with all of that, a given system may not remain identical when switching between modes. Let alone be identicle to any other system in different locations around the world.
Yet, the subjective character difference between OS and NOS seems to persist in my experience, even through changes to the system in which they are being evaluated. This is logical, else so many different listeners, via so many different systems would not prefer the sound of NOS playback. Obviously, I'm not suggesting that the sound of our individual systems systems must be, or even can be, identical. After all, how could they. I'm suggesting that all that is necessary, for what we are investigating, is that a given system be capable of revealing an audible difference between OS playback and NOS playback, not which sounds better on that system, or to the ears of that listener.
Last edited:
But, as said before, the staircase (i.e. ZOH) already brings in some attenuation, not only at 20 kHz, but, due to the overlayed sinc(x)-amplitude weighting, for the mirrored images as well.
For a 44.1 kHz NOS, the attenuation at the Nyquist-Frequency is 3.92 dB, a quite deep null exists at Fs and the multiples of Fs; the attenuation below and above Fs (and its multiples) is a bit asymmetrical and gets higher at images at higher frequencies. But it is at -14/16 dB around Fs for the bigger part of the images (means where the higher amplitudes occur for "normal" music).
This attenuations presumably already helps during the reproduction as otherwise the IMD would be too large. (Assuming that the reproduction electronics isn't already strongly bandlimited to 20 kHz)
A reasonable hypothesis would be that the subjective preference for NOS depends on the level of high frequency content in recordings.
Last edited:
If there would be a source that we all agree on thet can be trusted to give us what is on a recording (we all agree when we hear a real life violin that it is truely a violin right?) Then we do not have the "stress" of looking for equipment to cover things up or compensate for this. The concentrating on room acoustics and speaker placement is in order.
It is true that some recordings are bad and i have found this in an earlier mastering against a newer one to be true. It is nice to see however that lately the recording techiques are getting better and good remasters can be found.
It is true that some recordings are bad and i have found this in an earlier mastering against a newer one to be true. It is nice to see however that lately the recording techiques are getting better and good remasters can be found.
That is one thought, however, it is countered by the fact that some of us hear a clear difference between OS and NOS for midrange sounds, such as vocals, and even from bass range instruments. Those are sounds with no high range overtones.
What you are really suggesting, technically, is that the presence of the first image-band is responsible. That was one of our original hypothesis, however, Abaraxalito, and some builders of his 7th order analog filtered PhiDAC, report that the NOS character still remains. Further, studies show that acoustic attenuation of the ear is more than -50dB @ 20KHz for men over age 40. So, the presence of the image-bands don't appear to be responsible.
My comment was meant the other way round, means that a perceived advantage for NOS could be outweighed if the IMDs caused by the reproduction of inband high-frequencies and image-frequencies gets to unpleasant, as the lower sidebands will fall into a sensitive region.
