I tend to prefer slow roll offs and even the NOS mode - what does this say about my setup and preference? (Conclusion: back to fast linear)

[AnalogSteph]

Indeed, IMD is also increased:

Oh wow, the ADI-2 Pro is not a happy camper in NOS mode. The spray of low-level IMD products even within the audible range suggests that the analog stages may be slightly unhappy with the abuse they're subjected to (I agree a multitone isn't likely to look too pretty), and look at that noise floor modulation - it's easily up by 6 dB in the 19/20 kHz IMD test! That has to be something in the DAC itself.

NOS mode is essentially bypassing most of the digital filter so its lower group delay should come as no surprise. The SD LD (short delay low dispersion) filter appears to be a hybrid FIR + IIR filter, not a major surprise for a short group delay job. Its group delay spec is 10 samples with a deviation of +/-0.035 samples according to the AK4493 datasheet, I imagine the difference seen here would be a majority of that.

But doesn't that transient look much nicer on the NOS?

You're not listening with your eyes, are you? Your ears are more of a frequency domain device, and severely band-limited themselves.

 

[fieldcar]

Again, might be placebo, but now listening attentivly I think there is some hiss in the upper highs. Could be just be normal highs "enhanced" by my focussing. But nevertheless, it's not uncomfortable. Sounds kind of like even harmonics. Damn, I feel so weird for writing such unscientific stuff right now.
I also wanted to follow Amirs suggestion regarding roll offs, that why I ran the Gustard on the Fast filters for a while but there was a very small "hiss" in the audio band that is pretty much gone with slow filtering or NOS.

I wouldn't worry about not being scientific as most of us don't have access to an AP555. There is a possibility of the way the filter was implemented is faulty, and you're just using the resources you have available. My D50S filters all sound completely identical, even when listening to pink noise and sweeps. I score above average well at the klippel listening test (-39dB) and can hear up to about 16KHz. What's the best you've scored?

https://www.klippel.de/listeningtest/

 

[anphex]

Thank you for sharing this. Tried it just now. I just chose the first stimulus it offered me.

 

[anphex]

You're not listening with your eyes, are you?

If we go by this statement you just made all graphs on ASR pointless . But I get what you mean and appreciate your input, then again why does NOS sound better for me? Even before seeing all those graphs to placebo my mind. Considering that data at hand it should be worse or no difference at all. I know it's useless to describe sound but it appears as if the whole audio band I can hear is cleaned up. Nothing mushy, nothing hissy.

I'd love to see an in depth review of NOS my Amir some day. Too bad he didn't measure the NOS filter mode in the Gustard X16.

Todays AB-preference test was about 3/4 for the NOS, competing against M-Slow without seing the display and "randomizing" with hitting the remote wildly. Will try again tomorrow with fresh ears.

 

[voodooless]

Todays AB-preference test was about 3/4 for the NOS, competing against M-Slow without seing the display and "randomizing" with hitting the remote wildly. Will try again tomorrow with fresh ears.

Do you know if the NOS and OS modes deliver the same output level? If not, already this difference alone might explain what you think you hear. Only a proper controlled test can give better insight.

Or never mind that:

The thing is 25 dB down at 20 kHz, already -7 at 10 kHz. That is bound to sound different.

Just put the DAC is OS mode, get some EQ and pull down 20 kHz to about -25 dB.. see if it sounds about the same.

 

[anphex]

Do you know if the NOS and OS modes deliver the same output level? If not, already this difference alone might explain what you think you hear. Only a proper controlled test can give better insight.

You mean like NOS gives 0,5db more output and hence I prefer it naturally? This might very well be the case but I can't ascertain it. I can only claim that I think there isn't a difference, and that isn't worth much like you already said :/. Until this isn't explored further scientifically I just can go by "I like it better".

 

[fieldcar]

I'm glad to hear you found it useful. When I get -39dB, I'm using IEM's (ER4XR's or thieaudio Legacy 4's), but around -33dB when using my work speakers (Razer Nommo's). I find it maddeningly difficult to detect distortion at the end of the test, as the warbling of the voice and instruments disappear and it's only very minor defects that remain. The test does a really good illustration of how insensitive our ears are to distortion compared to DACs and amps that measure in the 100+dB SINAD range. Since we have such great measuring devices these days, I find that I'm only really sensitive to lossy Bluetooth codecs like SBC and AptX, especially with solo piano tracks, where the distortion sounds 'grainy' to my ears. (I know, subjective term BAD)

 

[Soniclife]

Or never mind that:

The thing is 25 dB down at 20 kHz, already -7 at 10 kHz. That is bound to sound different.

I use a much milder EQ boost from 10khz, and it's very audible, that is quite some EQ it's adding to the signal.

 

[anphex]

I'm glad to hear you found it useful. When I get -39dB, I'm using IEM's (ER4XR's or thieaudio Legacy 4's), but around -33dB when using my work speakers (Razer Nommo's). I find it maddeningly difficult to detect distortion at the end of the test, as the warbling of the voice and instruments disappear and it's only very minor defects that remain. The test does a really good illustration of how insensitive our ears are to distortion compared to DACs and amps that measure in the 100+dB SINAD range. Since we have such great measuring devices these days, I find that I'm only really sensitive to lossy Bluetooth codecs like SBC and AptX, especially with solo piano tracks, where the distortion sounds 'grainy' to my ears. (I know, subjective term BAD)

I found the test weird since it gave me really no calibration suggestion or the like. Since it's pretty late I didn't ramp up the volume. If I've listened a 85db I could probably dig a little deeper.

 

[anphex]

Do you know if the NOS and OS modes deliver the same output level? If not, already this difference alone might explain what you think you hear. Only a proper controlled test can give better insight.

Or never mind that:
View attachment 143921
The thing is 25 dB down at 20 kHz, already -7 at 10 kHz. That is bound to sound different.

Just put the DAC is OS mode, get some EQ and pull down 20 kHz to about -25 dB.. see if it sounds about the same.

Oh I saw the graph just now. Did you edit this? Nonetheless I find it hard to believe that in NOS it's already supposed to be down about 10db at 12khz but I will try it tomorrow with APO.

 

[voodooless]

Did you edit this?

Yes, I’m evil that way Sue me

 

[anphex]

As we say in Germany: Anzeige ist raus.
Just kidding.
Anyway, can't really sleep, so I played around a little with my APO config and added a high shelf with the same Q and starting point as the measurement you posted so it matches up and turned it on and off comparing all the while the all OS and NOS. But it's different. While the shelving does what it's supposed to the shelfing that is supposed to come with NOS isn't there. Such a huge drop in highs would surely be at least noticable. Instead I have my sound pretty linear as usual. I even played around with a sine generator in that nasty frequency range and didn't notice any drop compared to OS modes. Or maybe my ears aren't as good as I think and/or im a just retarded ... or actually sleepy.
Edit: I kept swapping back and forth between NOS and OS but there is absolutely no change that would indicate what has been measured up there.

 

[voodooless]

Curious! So you do notice with the EQ, but not in NOS mode?

Do you by any chance have an audio interface with an ADC?

 

[anphex]

Curious! So you do notice with the EQ, but not in NOS mode?

Do you by any chance have an audio interface with an ADC?

I have a not so old Steinberg UR22 MK2 24-Bit / 192 kHz interface laying around with two balanced inputs. Tell me what I should do and I could try out a few things

 

[voodooless]

I have a not so old Steinberg UR22 MK2 24-Bit / 192 kHz interface laying around with two balanced inputs. Tell me what I should do and I could try out a few things

Good enough! First would be to fire up REW and do some loop measurements, so X16 out to UR22 in. X16 at 44.1, UR at 192. If REW doesn’t let you do that, just try some other tools? Even playing some white noise should do the trick. That should give us an idea of the frequency response in NOS mode.

 

[AnalogSteph]

RMAA could definitely do it since it supports "offline" playback and recording. Just use e.g. Audacity (high quality settings: resampler - best quality) to resample a generated 24/192 test signal to 24/44. Then you can play that with whatever player on the DAC while RMAA is recording on the Steinberg. Mind you, playing back some 44.1 kHz white noise while using REW in RTA mode to show the spectrum should work along the same lines.

Just be warned that the ADC used in the UR22 (a CS4270 codec) has a pretty suspect antialias filter in quad speed mode, so don't be too surprised if you start seeing some funny business above 0.3 fs or so (~58 kHz). Should do for this though. They are basically skipping part of the digital filter. You can call it cheating, but the same design is used on several other Cirrus consumer-level ADCs (CS5340-42, CS5345-46). The analog input stage is generally expected to provide some additional filtering to help the ADC out.

If the response does turn out like in the X16 though, that filter is clearly intended for quad speed playback minimum (i.e. 176.4/192k). Makes sense since the point of "NOS" modes generally is using external upsampling, e.g. in software (the SoX resampler plugin for Foobar2000 would work fine, for example).

 

[anphex]

I never tried REW, so if one of you could make me a brainfriendly step by step guide I'd be very happy. If not I'd just click some buttons and adjust settings to stuff that looks reasonable and you're at the mercy of my experimenting. I will throw myself at it then after work today.

 

[voodooless]

Not exactly sure how it can be done in REW. In the preferences, I can only seem to select a single sample rate, and what we want is different sample rates for each device. RMAA should be able to do it as well and is simple to use as @AnalogSteph mentioned.

 

[Chrispy]

You suffer from OCD?

 

[Veri]

View attachment 143921
The thing is 25 dB down at 20 kHz, already -7 at 10 kHz. That is bound to sound different.

Just put the DAC is OS mode, get some EQ and pull down 20 kHz to about -25 dB.. see if it sounds about the same.

Note that the image you linked is NOT frequency response, FR is NOT down 25dB at 20kHz. The image shows FFT spectrum of white noise, to show how the filter works.

You can see the effect of NOS in the RME ADI-2 here https://www.audiosciencereview.com/...ac-and-headphone-amp-review.13379/post-850429
Treble does roll off early, -1dB down at 10kHz, -4dB at 21kHz. Not 25 dB down. The Slow filter actually rolls off harder, about 8dB at 21kHz.