New 28-bit DAC coming out.

[RickS]

Folks, let's scale back the off topic environmental discussion here please.

Thanks!

 

[Hemi-Demon]

You're missing the fundamental difference between single-path and multi-path D-to-A conversion. When program levels drop into the low-audible range --- where we best perceive the "atmospherics" of a room signature / reverb tail / subtle timbre --- single-path DAC THD ratio becomes at least two orders of magnitude worse than when measured at full-scale.

In multi-path architecture, at these same atmospheric levels, we convert using a different DAC core that is operating at full-scale (MSB), where its THD+N ratio is vastly better than any single-path architecture at these lower levels (all DACs today are single-path). This is perceived as a marked improvement in spatial and timbre realism (due to lower THD+N ratios).

Legendary mastering engineer Bob Katz --- an audio skeptic --- did a series of objective measurements, comparing THD of single-path vs. multi-path at low perceptual levels. I encourage you to read his results: https://positive-feedback.com/reviews/hardware-reviews/imersiv-d1-dac/

The imersiv website is now live, and the FAQ is in development. I'll check in to answer any questions. https://imersiv.com/faq/

It is hard to take something like this seriously when the marketing includes the obligatory "my wife" review

I asked my wife, Thea, to be there for the listening test. Over a 25-year period, I produced around 600 classical music / orchestral recordings in Northern California, mostly for public radio. The first thing I put on was a Handel recording for Delos – a large ensemble with choir, orchestra, and vocal soloists.
I had, of course, heard this recording hundreds of times in post-production. But hearing it through the multi-path DAC was a shock. I remember listening for about a minute then turning to Thea and saying “do you hear that?!” She nodded in agreement, and said something like “the music is 3D.”

 

[EJ3]

It is hard to take something like this seriously when the marketing includes the obligatory "my wife" review

But it's so CUTE when they do that (NOT!)

 

[signalpath]

But it's so CUTE when they do that (NOT!)

I agree. Have rewritten. Thanks for the constructive feedback. The story is 100% true, but probably nobody will believe it.
Please chime in with any other editorial suggestions. The website just went live.

 

[EJ3]

I agree. Have rewritten. Thanks for the constructive feedback. The story is 100% true, but probably nobody will believe it.
Please chime in with any other editorial suggestions. The website just went live.

It's just because it is reminiscent of many jokes involving disagreeing life long partners that, when, even if true, it borders on the not believable.

 

[popej]

The story is 100% true, but probably nobody will believe it.

It is certainly believable. But it shows that the person telling it doesn't understand the reasons behind the event. Which in turn raises doubts about their other claims.
Look for "Clever Hans effect".

 

[GXAlan]

@signalpath You should send a unit to @amirm to test. When an end user sends something in, measurements just get published. When a manufacturer sends it in, you get the opportunity to discuss findings or talk about the way you think the measurements will best be shown.

Imersiv D1 DAC - Positive Feedback

Imersiv D1 DAC

positive-feedback.com

I think the good example is this. With a -90 dBFS signal, or

-90 dBFS is 0.000123V. This is where you see that as the signal gets lower on the X-axis, noise is a bigger issue.

If we look at the -70 dBFS signal, it’s still lower than what Amir measures, the THD+N is 74 or 80 dB SINAD depending which number you are looking at. You’d guess the RME to be around 50 dB SINAD.

I think the audibility of this is a completely separate question, and I am the worst customer for this since I use Meyer Sound PA speakers at home, which have a lot of noise from being a super high-gain amplification structure — but the measurements are interesting for sure, if your priority ISN’T the 4V SINAD but the SINAD at lower levels.

 

[signalpath]

I think the audibility of this is a completely separate question ... the measurements are interesting for sure, if your priority ISN’T the 4V SINAD but the SINAD at lower levels.

Spot on. Multi-path performance is similar to excellent single-path performance at the highest measurable levels (see Katz paper). It's in the lower perceptual levels where multi-path improves THD+N. Often significantly. The improvement is easily measurable (see Katz paper). But more importantly, in our experience and in the experience of scores of Grammy and platinum engineers, the difference is immediately audible as improvement to both imaging and timbre realism. https://imersiv.com/user-comments/

This creates a bit of a paradox for those who assume that "if it measures the same, it will sound the same." We now have a new DAC does not measure the same. This unique DAC topology shows a measured improvement over all other DACs made today. This improvement occurs well within the perceptual range (-40dBFS), which is where we are most sensitive to subtle ambience and atmospheric cues (see Fielder papers, etc).

If we significantly reduce the THD+N at these lower perceptual levels, it's reasonable that highly trained ears will hear a difference, an improvement. But I don't think there's any way to "prove" that last statement. Each person would need to listen to a multi-path DAC in their own monitoring path and make a determination. We're blazing a new trail here and learning as we go. The ASR community will be very helpful in this project.

I keep coming back to HDR photography. It's a perfect analogy. We can read about how multi-image photography improves low-level dynamic performance. But it's just an abstract concept until we actually experience it with our eyes. Then it's an "ah ha" moment. It's the same with multi-path audio. It's just an abstract concept until you hear it.

 

[Salt]

Science demands double blinded, randomised studies ... translated to ASR: blinded ABx.

 

[signalpath]

Science demands double blinded, randomised studies ... translated to ASR: blinded ABx.

Yes. We've been doing self-blinded A/B (not ABX) audio tests for 35 years. It's the basis of everything we design, and the foundation of over 50,000 audio channels shipped.

When we first connected the kludge DAC prototype (2018 I think) on our reference system, the difference was immediately obvious. We did go back and forth with our existing reference DAC at the time, but not fast-switched blind A/B. Didn't need to. We immediately knew we had something of great value to the audio industry. If the difference had been marginal or "7/10" A/B hit, I wouldn't have wasted 7 additional years getting this to market.

During production beta testing (2024), I was invited to a blind A/B test at a large mastering house in Nashville (part of a $5B record company). They put on a classical music piece with large dynamic swing and ample room ambience, aligned levels to within 0.05dB, and then used their Masalec monitor controller for fast A/B switching.

It was immediately obvious. One DAC was like "seeing" deeper into the sound stage, the other became "shallower". Reminded me of looking at one of those "hidden 3D stereogram" drawings where you keep staring at it until it finally "pops into view" and then everything has depth and dimensionality. The mastering engineer doing the test said "it's like going from 35mm to IMAX".

I've heard from a number of D-1 beta testers who did their own A/B comparisons, probably mostly not blind. Almost to a person, the same results. I'm sure there will be a number of ASR contributors doing similar tests over the next years. Looking forward to it!

 

[SSS]

Yes. We've been doing self-blinded A/B (not ABX) audio tests for 35 years. It's the basis of everything we design, and the foundation of over 50,000 audio channels shipped.

When we first connected the kludge DAC prototype (2018 I think) on our reference system, the difference was immediately obvious. We did go back and forth with our existing reference DAC at the time, but not fast-switched blind A/B. Didn't need to. We immediately knew we had something of great value to the audio industry. If the difference had been marginal or "7/10" A/B hit, I wouldn't have wasted 7 additional years getting this to market.

During production beta testing (2024), I was invited to a blind A/B test at a large mastering house in Nashville. They put on a classical music piece with large dynamic swing and ample room ambience, aligned levels to within 0.05dB, and then used their Masalec monitor controller for fast A/B switching.

It was immediately obvious. One DAC was like "seeing" deeper into the sound stage, the other became "shallower". Reminded me of looking at one of those "hidden 3D stereogram" drawings where you keep staring at it until it finally "pops into view" and then everything has depth and dimensionality. The mastering engineer doing the test said "it's like going from 35mm to IMAX".

I've heard from a number of D-1 beta testers who did their own A/B comparisons, probably mostly not blind. Almost to a person, the same results. I'm sure there will be a number of ASR contributors doing similar tests over the next years. Looking forward to it!

It is just marketing. Sound of D-1? Maybe the kludge DAC sounded probably different and not so good by the ears of listnerers, but I doubt that anybody could say which is the real sound. It is all subjective and has no science verification.

 

[masterhw]

Hahaha, it is just marketing. Sound of D-1? Maybe the kludge DAC sounded probably different and not so good by the ears of listnerers, but I doubt that anybody could say which is the real sound. It is all subjective and has no science verification.

I can certainly understand skepticism but this is getting into outright insulting territory.
Yes the gold standard is a peer reviewed, repeatable test published in a scientific journal. This is also an incredibly high bar, and I almost couldn't think of something less widely accessible.

A long time member who has significant first hand experience with this product claims, in no uncertain terms, that he and other trained listeners under controlled conditions are able to reliably differentiate. I too am reserving something for wider testing, but this is a serious claim and refusing to give even an inch unless it has gone through a rather inaccessible process is well intentioned but perhaps a bit obtuse.

And would such a test truly advance audio science? On one hand it would be useful to understand for certain that this new(ish) DAC architecture can achieve perceptible improvements, but also as mentioned by signalpath above it is already shown to yield improvements up to -40dBFS which is already known to be audible. Thus there is limited raw scientific value to such an experiment. So if what you're wanting out of this is consensus on whether product claims are true then perhaps that is an unfair expectation to put on the scientific community, and perhaps it belongs more in the realm of audio engineering. Or product reviewing.

 

[Salt]

Agreed, that 'Hahaha' is a most inappropriate comment.

 

[GXAlan]

One of the reasons I lost interest in ASR beyond responding to threads that I started, is that there are double standards.

Many on ASR value engineering achievements when they can be measured even though it is well beyond the point of audibility.

We criticize stuff when the stuff does not provide improvements in audibility

The worst tube amp in here may not sound any worse to a synthetic straight wire with gain under blind testing, and yet, no one here likes tubes for fidelity. Knowing that people struggle to hear the difference from a 22 dB SINAD amp, tells you that the vast majority of products do not have any audible benefit. If you fall into the “only if it’s audible” camp, 100% of the electronic reviews here are not at all interesting beyond the subjective ergonomic and setup.

Then there is a group of ASR enthusiasts that value measurements even if it’s inaudible due to engineering prowess. That’s fine. We buy art. We buy fashion. We vote with our wallet by buying engineering art.

Here, this DAC seems to offer several orders of magnitude performance improvements at -40 dBFS and lower. That’s really cool when verified by a third party, which it has been done. It might NOT have much improvement at 4V out compared to others, but that isn’t what they are advertising.

So if you are unhappy with the claims of audibility, we should still be happy with independent measurements putting SINAD at -90 and -70 dBFS measurements at never before seen levels of performance, right?

 

[SSS]

I can certainly understand skepticism but this is getting into outright insulting territory.
Yes the gold standard is a peer reviewed, repeatable test published in a scientific journal. This is also an incredibly high bar, and I almost couldn't think of something less widely accessible.

A long time member who has significant first hand experience with this product claims, in no uncertain terms, that he and other trained listeners under controlled conditions are able to reliably differentiate. I too am reserving something for wider testing, but this is a serious claim and refusing to give even an inch unless it has gone through a rather inaccessible process is well intentioned but perhaps a bit obtuse.

And would such a test truly advance audio science? On one hand it would be useful to understand for certain that this new(ish) DAC architecture can achieve perceptible improvements, but also as mentioned by signalpath above it is already shown to yield improvements up to -40dBFS which is already known to be audible. Thus there is limited raw scientific value to such an experiment. So if what you're wanting out of this is consensus on whether product claims are true then perhaps that is an unfair expectation to put on the scientific community, and perhaps it belongs more in the realm of audio engineering. Or product reviewing.

Wow, a long text upon my input. Yor were not mentioned, it was signalpath. Anyway, there may be sound differences between very good DACs but it still stay that even a double blind test can not tell the absolute truth. The listeners ARE subjective and even if the majority favours a certain DAC, a test with other listerners may have a different statistical result. ABX is meant to be a good way to compare equipment by many people. But I do not believe this. The sounds are slightly different, but which is the better or realistic one is still questionable because the original sound of the recording is in general unknown. It is the circle of uncertainty as Sean Olive stated it.

 

[masterhw]

Wow, a long text upon my input. Yor were not mentioned, it was signalpath. Anyway, there may be sound differences between very good DACs but it still stay that even a double blind test can not tell the absolute truth. The listeners ARE subjective and even if the majority favours a certain DAC, a test with other listerners may have a different statistical result. ABX is meant to be a good way to compare equipment by many people. But I do not believe this. The sounds are slightly different, but which is the better or realistic one is still questionable because the original sound of the recording is in general unknown. It is the circle of uncertainty as Sean Olive stated it.

I would suppose a recording engineer should have some sense of what a recording of an orchestra should sound like. They may have been in the room when it was performed. There are some differences in things which are not necessarily better, just different. It has been alleged that this is not one of them. Audio science is the attempt to understand a perceived, often subjective, and complex psychoacoustic phenomenon. But if differences between sounds yield a statistically significant majority preference, then that is valid audio science. And the claim is that that has been done here.

 

[Salt]

I would suppose a recording engineer should have some sense of what a recording of an orchestra should sound like. They may have been in the room when it was performed. There are some differences in things which are not necessarily better, just different. It has been alleged that this is not one of them. Audio science is the attempt to understand a perceived, often subjective, and complex psychoacoustic phenomenon. But if differences between sounds yield a statistically significant majority preference, then that is valid audio science. And the claim is that that has been done here.

The problem is, that DAC is widely declared 'solved' on ASR, since no audible difference could be measured in a wide range of tested items, no matter what price.
From this point of view some irritations might come up when suddenly new 'audible' aspects are presented.
Interesting approach, esp. for ASR.

 

[Tell]

I keep coming back to HDR photography. It's a perfect analogy. We can read about how multi-image photography improves low-level dynamic performance. But it's just an abstract concept until we actually experience it with our eyes. Then it's an "ah ha" moment. It's the same with multi-path audio. It's just an abstract concept until you hear it.

That right image should be compared to a very compressed music signal, just a way to try to tonemap a very high dynamic range image into a low dynamic range display since there really ain't any display out there that can display an image that contains all levels of brightness that our eyes can see without extreme clipping. While in audio we can quite easily play the full range of what our ears can hear so there is no need to do any trickery, especially with DACs since it really is a solved problem. So I don't understand why this topic is still alive?

 

[masterhw]

The problem is, that DAC is widely declared 'solved' on ASR, since no audible difference could be measured in a wide range of tested items, no matter what price.
From this point of view some irritations might come up when suddenly new 'audible' aspects are presented.
Interesting approach, esp. for ASR.

Yes, having read the MQA/QRONO thread I certainly came away with the impression that there are certain areas that are 'safe' to try and further engineer, and some which will have people mad as hornets. I assume it is an overcorrection for the time in the audio hobby before I joined when marketing fluff was allowed to run rampant. As a recent follower I don't totally get why its such blasphemy to suggest that some of the DACs, some of the time, aren't perfection incarnate. Especially when the recent CS43131 thread showed us some DACs respond differently to music than they do test tones (hopefully more the exception than the rule though), showing particularly weak performance in the ranges people are most likely to actually use the devices in.

 

[IVX]

It's in the lower perceptual levels where multi-path improves THD+N.

To be honest, reduced only noise, if more specifically, the harmonics level of your d1 is 10x times worse than ES9039Q2M($5/pcs) at -70dbfs. Can you tell us what DAC chips are inside D1?