SMSL SU-10 DAC Review

[xnor]

I can understand you guys fury about the format, but wouldn't it be fair to at least not be so completely one sided when talking about it? I have no plan to purchase MQA CDs but let's be devil advocate for a second. My understanding is that you can play MQA CDs on any regular Redbook CD Player. You can't play any "other" High Res formats. Isn't it the main point?

You can play regular CDs on any regular CD player too, and it will be higher fidelity too.
Why would anyone play a signal with lower fidelity on a regular CD player than a regular CD supports?

Besides, you can also rip regular CDs, encode them losslessly, store bit-perfect copies, stream them, transcode them (for portable/streaming), and do the same with "HD audio" releases sold in non-DRM/vendor locked-in formats (where you actually get full access to the data that you paid for).
You can get CD drives that give you bit-perfect rips for $20.


It's like trying to resell me mp3s where ~15% of the data is held for ransom, is encrypted and can only be decrypted by making an additional payment for a new device. And the main point is that I can still play it on my old mp3 player, with a portion of the data being played as random noise? What?

 

[PeteL]

You can play regular CDs on any regular CD player too, and it will be higher fidelity too.
Why would anyone play a signal with lower fidelity on a regular CD player than a regular CD supports?

Besides, you can also rip regular CDs, encode them losslessly, store bit-perfect copies, stream them, transcode them (for portable/streaming), and do the same with "HD audio" releases sold in non-DRM/vendor locked-in formats (where you actually get full access to the data that you paid for).
You can get CD drives that give you bit-perfect rips for $20.


It's like trying to resell me mp3s where ~15% of the data is held for ransom, is encrypted and can only be decrypted by making an additional payment for a new device. And the main point is that I can still play it on my old mp3 player, with a portion of the data being played as random noise? What?

Why would they have lower fidelity than regular 44/16? aren't they high res files compressed (lossy) to be enclosed in a 44/16 container that then get unfolded into High Res? What is the PCM pre MQA encoding, and what is the PCM after decoding if you have a DAC that decode MQA CDs?

 

[Marc v E]

I always enjoy the reviews here and especially one on a well measuring device.

Having said that, a minidsp flex or even a denon 3700 avr would probably be the best buy due to eq features and subwoofer integration.

Given that there are so few contenders for the flex, should we test more avrs?

 

[PeteL]

Interestingly volume goes up another 2 dB (or so) so you have a bit of headroom.

What does that mean? Another dB compared to what? Analog or digital?

 

[xnor]

Why would they have lower fidelity than regular 44/16? aren't they high res files compressed (lossy) to be enclosed in a 44/16 container that then get unfolded into High Res? What is the PCM pre MQA encoding, and what is the PCM after decoding if you have a DAC that decode MQA CDs?

From sampling theory we know that content above 22.05 kHz will alias ("fold back down") if sampled at 44.1 kHz. The only way to be able "unfold" that again is by destroying information in the 0 to 22.05 kHz band (which includes the audio band... where the actually audible stuff is stored) beforehand. MQA does this down to leaving only 13 bits of information intact. The freed up bits are used to store the aliased content and nothing else. Actually, this ultrasonic content goes through another lossy compression algorithm...

It's a spectacularly bad compression scheme. Even the "high-res" variant is pretty poor. It takes up more space than a "mid-res" FLAC file while delivering lower fidelity at the same time.

 

[srkbear]

Hold on. Let's stop after the first comma: at this point? Reality check: audibly transparent conversion under normal usage has been reached many years ago and is available for a few dollars nowadays.
If there had been audible differences up to this point then we could have relied on (blind) listening tests instead of measurements all along and those listening tests would have all cross-confirmed the measurements. There also would not be such a divide between scientifically minded and subjectivist audiophiles.

How is selling something that is advertised to measure better (and does so) at a premium snake oil, regardless of the fact that there are no audible differences?
Unless the manufacturer makes the claim that under normal listening conditions there are actually audible differences, I don't see how it would be.



Sampling rates and bit depth never were an issue. The loudness war (which could have been ended decades ago with a simple solution), bad recording, bad mixing, bad mastering and bad musicians/singers are.
While 48/96 kHz do have some advantages, the industry obviously pushed "HD audio" to be able to resell you "HD ready" components and "HD music".

Currently, 13/17 bit lossy MQA that doesn't even compress better than lossless codecs at equivalent bit depth is being pushed, so that manufacturers can resell you "MQA ready" components and services/labels can resell you the same songs in a mangled form. And people fall for it, happily pay the extra fee.


What will the future be? Are you asking for something that actually makes a difference like object-based audio or the next MQA scam?

I’m well aware that we achieved the limits of human hearing well before now. This is the first time I’ve heard Amir state that we’ve hit the limits of his measuring rig, however. MQA doesn’t really seem germane to my question because it was marketed to solve a non-problem with streaming bandwidth, not with DAC performance, and with the aid of Stereophile and other industry-funded shills somehow got mistaken for an advance instead of a licensing scam.

I was sincere about what’s next in store for the DAC industry, and digital audio in general, if this latest SMSL offering has surpassed the reaches of the latest measuring rigs. I was also sincere in questioning SMSL’s tactics in charging an extra $200 for dual DAC chips, when Gustard and Topping can achieve statistically equivalent results from a single chip. This did not seem as much as a feat of engineering as it did a gimmick, whereas I see nothing insincere about the d90se, or the x18 for that matter, which is even more reasonably priced.

I chose to put my money on the d90se even though I could have achieved the same end result with cheaper DACs that didn’t “perform” as well, out of respect for Topping’s achievement—and the fact that such an enthusiastic audience exists here for these reviews suggests that others share the same values. Your response was lovely.

 

[HE-dbtr]

This is a very nice device and could’ be a replacement for my older Benchmark DAC2D. But there is one thing: Headphone output is missing.
A separate headphone amp is not that part i‚m looking for. But on the other hand: the benchmark is sounding great. The data are a little bit weaker but who cares?

I would keep your DAC2D. I don't think the SMSL is an upgrade unless you need Bluetooth or your DAC2D is not functioning well.

 

[PeteL]

From sampling theory we know that content above 22.05 kHz will alias ("fold back down") if sampled at 44.1 kHz. The only way to be able "unfold" that again is by destroying information in the 0 to 22.05 kHz band beforehand. MQA does this down to leaving only 13 bits of information intact. The freed up bits are used to store the aliased content and nothing else.

It's a spectacularly bad idea. Even the "high-res" variant is pretty poor. It takes up more space than a "mid-res" FLAC file while delivering lower fidelity at the same time.

You seem to know exactly the encoding algorythm, we all know that it's lossy but what is in the freed up bit and how the decoder recovers it in not public domain as far as I know. In all case it is a quite strong statement that you come up with, I would wish you would back it up with something more than just basic general knowledge of sampling theory. Encoding is not necessarily resampling, FLAC demonstrate this already. In all case, you really feel that just based on that you state today that, on Tidal, which has both MQA and 44/16 flac for many albums, that all the MQA files are rigorously of less fidelity than their redbook counter part? It is easy to demonstrate that you would add some artefacts compared to the High Res untouched one, but your own statement I have not heard many demonstration of that really, hoping you'd give me some reading material/ sources to back that up.

 

[mctron]

PeteL, what is it that you are actually arguing for/against? MQA is good?

 

[Trell]

PeteL, what is it that you are actually arguing for/against? MQA is good?

MQA is quite simply a scheme to extract money from every part of the audio chain. It has no audible benefits at best, and a degradation at worst.

This has been rehashed many times, tough, but people are free to spend their own money as they like.

 

[PeteL]

PeteL, what is it that you are actually arguing for/against? MQA is good?

I am not arguing, I am asking questions. I can only say that there are some really good sounding MQA files on Tidal. Majority of the time better than their 44.1 counterparts but it may be artificially induced, I don't know. It's demonstrated that it, obviously since it's lossy, will induce degradation compared to an equivalent High Res file. I want to know if it is demonstrated that it is also of less fidelity than Standard resolution files and why so.
In all cases High Res has never been about reproducing ultrasonics, In my book it should be about easier and smoother filtering outside of the audible range, thus reducing less phase and ringing arthefacts. That is what I believe MQA stands for but If it does none of that, it deserve some explanation.

And if it does that but it had nothing to the fidelity, I am ok with this, but let be consequent, and Also expose SMSL, Topping or any other manufacturers that offers a choice of filters as Snake oil sellers. If filtering don't matter, it don't matter there neither and there should be no benefits of one over the other.

 

[amirm]

Folks, these are desktop products. They are aimed mostly to headphone listeners which is a far larger market than home stereo. Headphones can be incredibly sensitive and block outside noise. This makes the requirements for dacs much tougher.

 

[solderdude]

Consider the filter choice is something the DAC chip manufacturer offers and DAC device manufacturers can make a decision to pick one, two, a few or all available filters.
More choice for the owner, regardless if audible or not.

 

[RandomEar]

I am not arguing, I am asking questions. I can only say that there are some really good sounding MQA files on Tidal. Majority of the time better than their 44.1 counterparts but it may be artificially induced, I don't know. It's demonstrated that it, obviously since it's lossy, will induce degradation compared to an equivalent High Res file. I want to know if it is demonstrated that it is also of less fidelity than Standard resolution files and why so.
In all cases High Res has never been about reproducing ultrasonics, In my book it should be about easier and smoother filtering outside of the audible range, thus reducing less phase and ringing arthefacts. That is what I believe MQA stands for but If it does none of that, it deserve some explanation.

And if it does that but it had nothing to the fidelity, I am ok with this, but let be consequent, and Also expose SMSL, Topping or any other manufacturers that offers a choice of filters as Snake oil sellers. If filtering don't matter, it don't matter there neither and there should be no benefits of one over the other.

This thread really isn't about MQA, but well... You can watch Amir's video about an example MQA file. The file clearly shows that MQA raises the noise floor above 15 kHz and - in this range - seems to encode information used to reconstruct ultrasonic stuff. That clearly demonstrates, that information in that (audible) range is lost, albeit at a very low level. As the MQA compression has been shown to be lossy, that information can't be fully recovered by "unfolding" the file. Equally, the reconstructed ultrasonic content is incomplete and lossy.

The advantage of "smoother filtering" you might get by using HighRes files can equally be achieved by upsampling, without destroying parts of the audible information by using MQA. I further haven't seen any controlled study that proves the audibility of the "phase and ringing artefacts" you mention.

The fact that some MQA files on Tidal might sound better to you or me is most likely explained by the fact that they're using a different master. The same master could just as well be delivered in 44.1 kHz / 16 bit without a lossy compression.

 

[Toni Mas]

Consider the filter choice is something the DAC chip manufacturer offers and DAC device manufacturers can make to decision to pick one, two, a few or all available filters.
More choice for the owner, regardless if audible or not.

Imho, with 44/48khz material sharp filters make sense, but upsampling at the máximum possible frequency and use a slow or no filter at all sounds better for me. At least with my cheap and already almost outdated Topping E30.

 

[TonyJZX]

I actually literally spat out "HOLEE SCHITT" (not the company) when I saw the graphs... surprised and yet.... NOT

we're accustomed to amazing stats from these guys... they're in an arms race but I would echo what their long term goals are

I cannot comment on how the local population is serviced by these products. I would question thought what peoples' expectations are of a $900 product once outside of the 2yr warranty? I expect this is largely a single board product. One would hope they could do reasonably priced board replacements? But given dacs are largely reliable but one could say that about digital amps and yet, here we are.

Once in the past I bought Marantz's top of the line THX amp and they too werent too happy to help me outside of warranty. In the antipodes they cant supply replacment boards. And so I got a elderly Chinese electronics engineer to fix it. And it still works! But one cant expect this on a state of the art dac.

 

[Doodski]

But given dacs are largely reliable but one could say that about digital amps and yet, here we are.

I guess you are referring to class D amps. They are analogue amps and not digital amps.

 

[pseudoid]

...They are aimed mostly to headphone listeners which is a far larger market than home stereo...

I am really shocked at this fact!
Even if these headphone users are 50/50 split between home and mobile/portable listeners.

 

[Toni Mas]

I am really shocked at this fact!
Even if these headphone users are 50/50 split between home and mobile/portable listeners.

Sad but true... Metaverso early users...

 

[PeteL]

This thread really isn't about MQA, but well... You can watch Amir's video about an example MQA file. The file clearly shows that MQA raises the noise floor above 15 kHz and - in this range - seems to encode information used to reconstruct ultrasonic stuff. That clearly demonstrates, that information in that (audible) range is lost, albeit at a very low level. As the MQA compression has been shown to be lossy, that information can't be fully recovered by "unfolding" the file. Equally, the reconstructed ultrasonic content is incomplete and lossy.

The advantage of "smoother filtering" you might get by using HighRes files can equally be achieved by upsampling, without destroying parts of the audible information by using MQA. I further haven't seen any controlled study that proves the audibility of the "phase and ringing artefacts" you mention.

The fact that some MQA files on Tidal might sound better to you or me is most likely explained by the fact that they're using a different master. The same master could just as well be delivered in 44.1 kHz / 16 bit without a lossy compression.

I didn't dispute any of that, of course it can be achieved by upsampling, but why care about the filters that are presented to us in this review? Clearly this DAC does not upsample.or if it does these are filters designed to get a stop band at 44.1 for a 44.1 file. For your rise of noise floor in the video at 16k and up. It never pass -100dB.... CD's dynamic range is 93 dB after dither.