New 28-bit DAC coming out.

[RandomEar]

Take a look at the system used in that review... Gauder, Gryphon, Grimm, Perlisten. I’ll give the benefit of the doubt that in a system like that, the experience described is real.

Due to the placebo effect, maybe. This audiophile idea that spending tens of thousands on a rare device made by some niche manufacturer must make it sound better will always be absurd. Bigger manufacturers invest millions of dollars more into good R&D and design and manufacture better performing devices. That's just how it is.

That Gryphon amp for example takes nearly an hour (!) to warm up and get THD into acceptable territory - and even then, it's around -77 dB. This is 60s-level technology. You can easily get better than -100 dB today, for a 25th of the price of that amp. It's expensive and looks cool, but its engineering sucks and it performs like something from 60 years ago.

And listening to a DAC with that kind of dynamic range on speakers is just idiotic. You'll loose 10-20 dB of possible dynamic range due to masking by the room noise floor (around 25-35 dB). You'll never need the higher dynamic range of the D-1 for a multitude of reasons, but not testing it with the best closed back headphones or IEMs you've got just shows that the reviewers did not know what they were doing. You absolutely need the isolation those provide to have any chance of hearing any theortical advantage the DAC might give you.

The only scenario where something like the Imersiv DAC could be audibly advantageous is in an ultra-quiet, highly resolving, DSP-free 2.0 system with excellent speakers and ideal recordings, where the system noise floor is low enough to expose extremely subtle low-level differences.

No. As explained above, any system with speakers is precisely not where the advantage might show, because your room noise floor will be limiting you. You're limited in peak SPL to about the pain threshold (120-130 dB), so you need the environmental noise to be at or below 0 dB to even gain the theoretical possibility of hearing any difference to other DACs.

Even then, those differences would be very small and difficult to isolate without controlled, level-matched comparisons.

Outside of that, speakers, room, and processing dominate the audible result.

The differences will not exist, because the music you listen to has a significantly higher noise floor and lower dynamic range than the D-1. Likely about an order of magnitude worse. There's no ADC with even remotely similar performance regarding noise today, so you'll not be able to hear any advantage with recorded music until that changes.

 

[IVX]

This would mean:
- quality control not possible,
- no real life gear, that would profit from better specification.

Interesting that APx555 can't measure the Cirrus DAC(CS43131/43198) either. Cirrus officially offered some LNA circuitry to add before APx555 to get -130db(A)@2Vrms

 

[knownothing]

It controls the headroom *and* the minimum difference in sound level that can be captured (resolved) -- aka the resolution. I've seen it compared to 'grain' in the video realm.


Sample rate does not determine the resolution in that sense. It determines the highest frequency that can be captured.

Thanks for weighing in.

In addition to the Nyquist Theorum, my understanding is that higher sample rates (e.g., 96 kHz) can provide more precise, detailed sound capturing.

For example, my understanding is that higher sample rates take more "snapshots" per second (e.g., 96,000 vs. 44,100), leading to a more precise digital representation of the original analog wave's shape. Because of the temporal sensitivity of human hearing, more accurate capture or reconstruction of musical sign wave can be perceived as better representation of tone and timbre in reproduced music.

Also high sample rates prevent aliasing, where high frequencies "fold back" into the audible range as incorrect lower-frequency noise.

I am wondering if the Imersiv D1 also applies any unique engineering to address this issue, or uses fairly standard approaches with all of the new tricks in the area of how it handles bits, loudness and noise.

The reason I bring this up is 1) the Imersiv D1 designer claims improvements in tone and timbre with his multi path design, and 2) I have four delta sigma DACs and two fpga Chord DACs and think especially the ESS DACs have what I would call showy and “atmospheric” treble while the heavily upsampling Chords are not quite as airy sounding but nail the tone and dynamics of voices, pianos and trumpets, and make redbook and even mp3/4 recordings sound less compromised.

I am wondering if the Imersiv design can provide both ultra low noise at low signal levels and excellent handling of sample rates and time domain. Perhaps this is a question better posed to @signalpath.

kn

 

[Bland_username42069]

Even at 24 bits there is no blackness barrier (Assuming I understand what you mean by that). The noise floor of 24 bit is at least 40dB below what the most sensitive human ear can detect. It is also around 50dB lower than the noise floor of your typical amp, and 24dB below the noise floor of the best amps ever tested here.

It was just an extremely low grade attempt at Stereophile parody.

 

[markus]

For example, my understanding is that higher sample rates take more "snapshots" per second (e.g., 96,000 vs. 44,100), leading to a more precise digital representation of the original analog wave's shape.

In the context of the limits of human hearing, the answer is no. This is precisely what the Nyquist–Shannon sampling theorem explains.

 

[knownothing]

In the context of the limits of human hearing, the answer is no. This is precisely what the Nyquist–Shannon sampling theorem explains.

My understanding is that the sampling rate of 44.1 kHz was inherited from PCM adaptors which was the most affordable way to transfer data from the recording studio to the CD manufacturer at the time the CD specification was being developed.

The Nyquist–Shannon sampling theorem says the sampling frequency must be greater than twice the maximum frequency one wishes to reproduce. To capture the human hearing range of roughly 20 Hz to 20,000 Hz, the sampling rate had to be greater than 40 kHz. OK, I get that.

While this theorem captures the sampling rate needed the represent the full frequency spectrum related to human hearing - it does not treat any artifacts that might be perceived by an end listener due to “steps” in the sine wave between sample points - even at 44,100 samples per second. Why else would any designer consider up sampling a 44.1kHz program in a DAC if that level of resolution truly provides “perfect sound forever”?

What is the empirically determined limit of detection of human hearing with respect to sample rate? Or stated differently what is the time domain (not frequency domain) limitation of human hearing to decipher discontinuities in a sign wave?

This I think may be why some suggest with all its limitations, analog replay has superior tone, timbre and less listener fatigue compared to at least poorly implemented digital. Greater - infinite sample frequency is superior in some respects even if noise reduction and dynamic range are far superior for modest implementations of digital reproduction.

Perhaps I am missing something very important here…

kn

 

[RandomEar]

Thanks for weighing in.

In addition to the Nyquist Theorum, my understanding is that higher sample rates (e.g., 96 kHz) can provide more precise, detailed sound capturing.

For example, my understanding is that higher sample rates take more "snapshots" per second (e.g., 96,000 vs. 44,100), leading to a more precise digital representation of the original analog wave's shape. Because of the temporal sensitivity of human hearing, more accurate capture or reconstruction of musical sign wave can be perceived as better representation of tone and timbre in reproduced music.

"Temporal sensitivity" is just audiophile nonsense. 16 bit / 44.1 kHz gives you a time resolution of about 0.1 ns. Even if you assume that you lose an order of magnitude in resolution due to noise + dither, 1 ns is far, far better than anything humans could ever discern. As pointed out by @markus, higher sampling rates than 44.1 are not required to faithfully reproduce all frequencies in the audible range. I would accept that 48 kHz gives you a slightly more comfortable "buffer", but that's about it.

Also high sample rates prevent aliasing, where high frequencies "fold back" into the audible range as incorrect lower-frequency noise.

I am wondering if the Imersiv D1 also applies any unique engineering to address this issue, or uses fairly standard approaches with all of the new tricks in the area of how it handles bits, loudness and noise.

This is not an issue for modern DACs. First of all, it is the responsibility of the audio engineer capturing the sound to make sure that his recording chain is properly band-limited for the recording format he choses. If the engineer fails to do so, you can't correct this in post processing. Second, reconstruction filters are used in upsampling since oversampling DACs became a thing. They suppress ultrasonic content and images alike. There are no "new tricks" in this field, since the most faithful filter options are well known and have been implemented in DACs for decades.

The reason I bring this up is 1) the Imersiv D1 designer claims improvements in tone and timbre with his multi path design, an

He's free to prove his claims using proper level-matched, double blind listening tests. Until he does so, they are just claims, not facts.

2) I have four delta sigma DACs and two fpga Chord DACs and think especially the ESS DACs have what I would call showy and “atmospheric” treble while the heavily upsampling Chords are not quite as airy sounding but nail the tone and dynamics of voices, pianos and trumpets, and make redbook and even mp3/4 recordings sound less compromised.

Unless you selected vastly different filters (fast vs. slow roll-off or NOS) or other settings like EQ, it is extremely unlikely that there is any audible difference at all between any of your DACs. Sighted listening tests without level-matching and using long switching times will give you useless biased and non-reproducible results.

I am wondering if the Imersiv design can provide both ultra low noise at low signal levels and excellent handling of sample rates and time domain. Perhaps this is a question better posed to @signalpath.

kn

You are just repeating the marketing blubber used by manufacturers. There is nothing magic about DACs. Their engineering is well understood and most good modern implementations are audibly transparent - as they should be.

 

[RandomEar]

My understanding is that the sampling rate of 44.1 kHz was inherited from PCM adaptors which was the most affordable way to transfer data from the recording studio to the CD manufacturer at the time the CD specification was being developed.

The Nyquist–Shannon sampling theorem says the sampling frequency must be greater than twice the maximum frequency one wishes to reproduce. To capture the human hearing range of roughly 20 Hz to 20,000 Hz, the sampling rate had to be greater than 40 kHz. OK, I get that.

While this theorem captures the sampling rate needed the represent the full frequency spectrum related to human hearing - it does not treat any artifacts that might be perceived by an end listener due to “steps” in the sine wave between sample points - even at 44,100 samples per second.

There are no "steps" in the waveform. For the time between samples, the signal is undefined. The supposed steps will be smoothed/interpolated during upsampling in any half-competent DAC and are not part of the signal that you listen to.

Why else would any designer consider up sampling a 44.1kHz program in a DAC if that level of resolution truly provides “perfect sound forever”?

See this thread for an explanation of what upsampling and reconstruction filters do in practice. It is not about getting to higher sampling rates per se - when you convert a discrete signal into a continuous one, you need to define what happens between the discrete sample points of the original signal. This is handled by upsampling and reconstruction filters.

What is the empirically determined limit of detection of human hearing with respect to sample rate? Or stated differently what is the time domain (not frequency domain) limitation of human hearing to decipher discontinuities in a sign wave?

There is no "limit with respect to sampling rate". There are only frequency limits, level thresholds and so on. Time and frequency domain are 100% interchangeable. You can transform a signal from one domain to the other and back and you end up with exactly the same data. The engineers who design the DACs and everything else you use day to day work with this precise knowledge. This "but t-t-t-time domain!"-stuff is audiophile blubber. Engineers handling digital signals never talk about this shit.

This I think may be why some suggest with all its limitations, analog replay has superior tone, timbre and less listener fatigue compared to at least poorly implemented digital. Greater - infinite sample frequency is superior in some respects even if noise reduction and dynamic range are far superior for modest implementations of digital reproduction.

Perhaps I am missing something very important here…

kn

Most LPs you listen to have been digitally processed before being pressed onto vinyl. This idea that "digital sounds worse" or all the other silly ideas about "fatigue" and whatnot are absolute hore-s-h-i-t. There is no such thing and only people clueless about audio engineering try to perpetuate this stuff.

 

[Soniclife]

Perhaps I am missing something very important here

You are, sampling theory is not at all intuitive. Watch the video linked above.

 

[antcollinet]

The only scenario where something like the Imersiv DAC could be audibly advantageous is in an ultra-quiet, highly resolving, DSP-free 2.0 system with excellent speakers and ideal recordings, where the system noise floor is low enough to expose extremely subtle low-level differences.

Can it though. How so, when standard well designed dacs exhibit noise and distortion below the level any human ear can detect? You can't get "more inaudible" than inaudible.

 

[antcollinet]

It was just an extremely low grade attempt at Stereophile parody.

Ah - you got Poed

 

[thcdru2k]

Can it though. How so, when standard well designed dacs exhibit noise and distortion below the level any human ear can detect? You can't get "more inaudible" than inaudible.

Probably not ; until someone does some measurements/tests I just say could out of precaution. I've never had the pleasure of listening to such a system myself so cannot rule out the possibility it maybe could make a difference. Like with thousand dollar USB cables I would feel confident saying no difference. But with very nice equipment in a treated room with an expensive DAC I guess I'm willing to be more open minded.

 

[Purité Audio]

I have represented many expensive dacs ( before I knew better) it is quite a shock when you perform your first level matched unsighted comparison and you realise that the differences you previously perceived have disappeared.
Sadly a huge percentage of Hi-Fi is just a con.
Keith

 

[IPunchCholla]

Probably not ; until someone does some measurements/tests I just say could out of precaution. I've never had the pleasure of listening to such a system myself so cannot rule out the possibility it maybe could make a difference. Like with thousand dollar USB cables I would feel confident saying no difference. But with very nice equipment in a treated room with an expensive DAC I guess I'm willing to be more open minded.

An easier path to this is to test your hearing for the elements we know matter. There are online tests for distortion (klippel), dB sensitivity, and you can get a decent audio gram from several apps with the right headphones/IEMs. Not sure if there is a test for noise sensitivity.

It is also good to get a sense of how susceptible to bias you are. I’m not sure if there are tests for this, but I have accidentally tested myself by not knowing my sub was physically off and spending a bunch of time digitally adjusting output and hearing differences before realizing it. Or hearing massive differences between mixes that disappear when blind A/B tested. From those experiences I know my brain can generate over 3dB perceived differences that are not present and can’t reliably differentiate 2dB differences that are present. It’s part of being human.

 

[Mr. Widget]

I have represented many expensive dacs ( before I knew better) it is quite a shock when you perform your first level matched unsighted comparison and you realise that the differences you previously perceived have disappeared.

100%

Sadly a huge percentage of Hi-Fi is just a con.

I prefer to think of it as irrational exuberance.

 

[antcollinet]

An easier path to this is to test your hearing for the elements we know matter. There are online tests for distortion (klippel),

Indeed.

Just try this one @thcdru2k to find out what level of distortion you personally can detect in real music. And then realise (if your abilty conforms with the vast majority of people tested) that even inexpensive dacs are routinely achieving distortion 1/1000th of the level that you can hear.

Feel free to post a screenshot of your result:

Listening Test

www.klippel.de

 

[jan.didden]

You guys may want to review Stuart Yaniger's review of the Immersive DAC in the NOv 25 issue of AudioXpress. Hard measurements you so love instead of endless speculation!

audioXpress November 2025 Sets the Roadmap for Audio Signal Processing

Get the broader picture on DSP, AI, and Edge Processing with our latest Market Update, and try a hands-on project as Philipp P. Klose explains why and how he designed his headPro Audio Processor for headphone correction. Estefanía Cano explains everything about Music Style Transfer, Dani...

audioxpress.com

Jan

 

[antcollinet]

You guys may want to review Stuart Yaniger's review of the Immersive DAC in the NOv 25 issue of AudioXpress. Hard measurements you so love instead of endless speculation!

audioXpress November 2025 Sets the Roadmap for Audio Signal Processing

Get the broader picture on DSP, AI, and Edge Processing with our latest Market Update, and try a hands-on project as Philipp P. Klose explains why and how he designed his headPro Audio Processor for headphone correction. Estefanía Cano explains everything about Music Style Transfer, Dani...

audioxpress.com

Jan

Mostly we are not discussing measurements in this case, but audibility.

Do the measurements in that review somehow demonstrate audibility compared with a standard competent Dac? I would look for myself but unfortunately your link doesn't include them.

 

[RandomEar]

You guys may want to review Stuart Yaniger's review of the Immersive DAC in the NOv 25 issue of AudioXpress. Hard measurements you so love instead of endless speculation!

audioXpress November 2025 Sets the Roadmap for Audio Signal Processing

Get the broader picture on DSP, AI, and Edge Processing with our latest Market Update, and try a hands-on project as Philipp P. Klose explains why and how he designed his headPro Audio Processor for headphone correction. Estefanía Cano explains everything about Music Style Transfer, Dani...

audioxpress.com

Jan

Skimmed over the audioXpress article. The results are not surprising: The DAC measures well and seems to be well engineered. Distortion is state of the art, noise is better than the former state of the art for attenuated signals, just as expected. Frequency response, impulse response and filter implementations are all pretty standard. The linearity seems to be a tiny bit better at the lowest levels (<-120 dB) which Amir doesn't even measure anymore. The level switching between the low and the high path is implemented well with less than 0.02 dB of error on the worst channel. The delay is a bit higher than typical (either 2.1 or 16.2 ms, depending on the mode) and the DAC also offers a distortion mode with tunable levels of harmonic distortion - a nice feature, but it seems odd to me in a device striving for absolute audible perfection.

Now the most important results for our subjective rebels: The reviewer and three "musically inclined" students in their 20s could not detect any audible differences to an RME ADI 2 FS B, neither in sighted nor in blind testing (both level-matched). Nada. Nothing. Quelle surprise! The reviewer also challenged imersiv to provide evidence for their claim of passing blind tests "ten times out of ten".

EDIT: Compared to ADI 2 FS B, not the DAC FS.

 

[Soniclife]

You guys may want to review Stuart Yaniger's review of the Immersive DAC in the NOv 25 issue of AudioXpress. Hard measurements you so love instead of endless speculation!

audioXpress November 2025 Sets the Roadmap for Audio Signal Processing

Get the broader picture on DSP, AI, and Edge Processing with our latest Market Update, and try a hands-on project as Philipp P. Klose explains why and how he designed his headPro Audio Processor for headphone correction. Estefanía Cano explains everything about Music Style Transfer, Dani...

audioxpress.com

Jan

I'll look forward to them when they are no longer behind a pay wall, but I don't detect much cynicism about the engineering or measurements in this thread, just how audible the improvement is.