CHORD M-Scaler Review (Upsampler)

[AudioSceptic]

Is to extract $6000 from the customer's bank account and with Chord/ Watts' relentless marketing campaign, make them feel special about it.

But only if you have a good enough Chord DAC and other components to appreciate it, of course.

But it's claimed to improve the output from *any* DAC, even one from 1999 (see JA's Stereophile review).

 

[JSmith]

Cost of products isn't a direct relation to component parts - there's development time and costs and market size to account for in order to provide the required return on investment. I'm sure you know this.

Sure, but $35 worth of parts and maybe a $120 case for $5K+? Sorry, that doesn't wash with me... $400 - $500 would be more acceptable, if it actually did anything positive that is repeatable in controlled blind testing. Doesn't Chord make enough money to afford testing of their products, both measurements wise and controlled listening? C'mon man... this is a dud product, plain and simple.


JSmith

 

[Jimbob54]

This product is certainly not crap. It should be measured with Dave or tt2 or even qutest/mojo 2 at 16x upsampling to get that brick wall filter attenuation.

Have a look at Goldensounds measurements of the Dave with and without M-scaler

 

[Sokel]

Let me ask a silly question.
As far as I understand that thing connects to dac through spidf,right?
What kind of spidf plays more than 24/192 so it can receive the X16 resampling?
All the dacs I know only play 24/192 through spidf.

 

[Jomungur]

Selective quoting... you missed out the post where I stated that I had missed the 16x graph. My primary interest was why you did not try the other upscaling rates for your listening test.

I thought the 'glitch' was when going from 4x to 16x, going from memory? I seem to recall that it's related to switching from one cable to two cables? I cannot recall with certainty but I don't recall a glitch from 2x to 4x. I have a feeling that 2x and 4x have differences compared to 16x. I don't think there is a glitch going from 16x to bypass.

The tone of your response seems to be somewhat defensive. I did not question the entire test - I asked specific questions about apparent omissions (one of which I later realised had been done), which is not the same as questioning the entire test.

Obviously I don't have my M Scaler at the moment, but you can't go straight from bypass to 16x. You have to hit the OpX button three times from bypass. The first takes you to 2x, the second to 4x, the third to 16x. This is because of Chord's design; it's not a dial or just hitting one of 4 buttons, it's cycling a single button that changes color. You don't have to switch cables. On my DAVE, when I have the dual BNC hooked up it goes from 2x to 4x to 16x without having to disconnect anything. But I don't recall a way that you can jump straight from bypass to 16x, and there is a slight pause when changing Opx modes, or even if you hit the Video mode. You are right, though, that the longest pause is between 4x and 16x.

EDIT: reread your post, I see you weren't implying one could go straight from bypass to 16x. It does seem from everything I've gathered, including Watt's own comments, that Blumlein is right: the M Scaler was really about 16x and the 2x and 4x were throwaway features to accommodate non-Chord DACs and increase the market. It'll be hard to ever really know, but I think it's a useful educated guess for someone who is considering buying this and has a non-Chord DAC.

 

[maverickronin]

But secondly, and most importantly. Why is there latency?

Just because it uses linear phase FIR filters with a lot of taps.

 

[Jimbob54]

Let me ask a silly question.
As far as I understand that thing connects to dac through spidf,right?
What kind of spidf plays more than 24/192 so it can receive the X16 resampling?
All the dacs I know only play 24/192 through spidf.

Thats why you need the special dual bnc connections I believe.

 

[Sokel]

Thats why you need the special dual bnc connections I believe.

To double it?
Even so,you'll still be far from 24/705.6 which the X16 is.
Strange things.

 

[voodooless]

To double it?
Even so,you'll still be far from 24/705.6 which the X16 is.
Strange things.

I think one can squeeze 384 kHz out of a coaxial SPDIF interface. Double that to get the 768 kHz.

 

[Sokel]

I think one can squeeze 384 kHz out of a coaxial SPDIF interface. Double that to get the 768 kHz.

Maybe you're right but I have never come across a dac that can do more than 24/192 through spidf.
Moreover the x16 is useless to any third party dac without the special dual BNC so it should be a mention in the manual.

 

[Jomungur]

Maybe you're right but I have never come across a dac that can do more than 24/192 through spidf.
Moreover the x16 is useless to any third party dac without the special dual BNC so it should be a mention in the manual.

In fairness, Chord makes this clear even on the website that you can only get 16x with certain Chord DACs, including Hugo 2.

 

[Sokel]

In fairness, Chord makes this clear even on the website that you can only get 16x with certain Chord DACs, including Hugo 2.

That's good to know.

 

[Jomungur]

I suppose those that will buy the M Scaler would expect that Output B sounds better than Output A

Well, I suppose it's possible? The M Scaler would then be kind like an advanced DSP device.

 

[Dogcoop]

There is a major flaw in his argument. The content before it hits M-scaler is already dithered. M-scaler processes that data and to avoid quantization error, adds another layer of noise. In that regard, it degrades the signal to noise ratio of the system regardless of which form of dither it uses.

In other words, there is no free lunch here. The incurred loss of signal to noise ratio better be accompanied with higher fidelity which sadly, we don't have here.

That aside, his assertion is wrong about superiority of Gaussian noise relative to Triangular Dither. Here is the paper that is considered the bible of dither:
AES paper, DIGITAL DITHER 2412 (C-8)
Stanley P. Lipshitz
John Vanderkooy
Audio Research Group
University of Waterloo
Waterloo, Ontario,

View attachment 216824
As you see, triangular dither is superior in both performance and level of noise it adds to the system. This is why it has become the de facto standard in digital audio processing.

As you see at the bottom, the reason to use Gaussian is because it is simpler/cheaper to implement. Not that it is better. In a $6,000 device, I expect the better dither to be used, than the cheaper one that raises the noise floor yet again.

hi @amirm
i don’t know why Mr. Watts doesn’t reply to you directly on this site, but this was his post on another site regarding your post on dither. As my technical expertise is limited, could you evaluate his statement below and let me know if it makes any sense. Thnx!


If you actually look at the original dither paper:
Quantization and Dither: A Theoretical Survey*
STANLEY P. LIPSHITZ, AES Fellow,ROBERT A. WANNAMAKER, AND JOHN VANDERKOOY, AESFellow**

You will see it's much more complicated. So rectangular dither is rejected due to noise modulation; but triangular dither has errors too. Look down to the conclusions in the paper you will see:

"For audio signal processing purposes there seems to
be little point in rendering any moments of the total
error other than the first and second independent
input. Variations in higher moments are believed to be
inaudible, and this has been corroborated by a large
number of psychoacoustic tests conducted by the authors
and others [18], [23]."

Triangular PDF dither covers the first and second moments of the total error, but not higher moments - that means triangular dither will create errors that are signal dependent.

My own listening tests concerned comparing noise shaped outputs at 16FS when one can employ a suitable noise shaper to perfectly reproduce in-band audio signals (perfectly being defined as the ability to reproduce a -301dB signal with phase and amplitude fidelity to 0.001 db and 0.001 deg of phase) to dithered outputs. The rectangular dither gave the worst performance, with suppression of depth and detail resolution compared to the noise shaped outputs. Triangular was better, and psuedo Gaussian gave the closest to the noise shaped reference. But even with the psuedo Gaussian, there was still a noticeable degradation in depth and detail resolution performance compared to the noise shaped reference.

 

[AndrewC]

Thats why you need the special dual bnc connections I believe.

Chord's Dual-BNC is not special (proprietary), it's in fact compliant to the AES3-2009 specification ("dual-wire mode") - to carry two channels above 192kHz bit rates, and the AES-2id specification for the BNC interface.

 

[Jimbob54]

Chord's Dual-BNC is not special (proprietary), it's in fact compliant to the AES3-2009 specification ("dual-wire mode") - to carry two channels above 192kHz bit rates, and the AES-2id specification for the BNC interface.

And someone mentioned they supply with the device. I didnt mean to imply proprietary. Perhaps "uncommon" might be a better description.

 

[DavidEdwinAston]

Can I comment that I am content with my Qutest, and don't think it was mega expensive. (The odd situation of not being able to "see", what voltage it is using, and not receiving any info on this from Chord, is something to consider.)
I appreciate, that thanks to this site, when it fails, it's replacement can be obtained for less money!
Re the M scaler, my limited brain thought. "Digital music is a series of steps. Analogue a single line. Therefore, a device to make the steps smaller, must improve the sound".
Thank you Amir for showing me how wide of the mark, that reasoning was!

 

[Veri]

My own listening tests concerned comparing noise shaped outputs at 16FS when one can employ a suitable noise shaper to perfectly reproduce in-band audio signals (perfectly being defined as the ability to reproduce a -301dB signal with phase and amplitude fidelity to 0.001 db and 0.001 deg of phase) to dithered outputs. The rectangular dither gave the worst performance, with suppression of depth and detail resolution compared to the noise shaped outputs. Triangular was better, and psuedo Gaussian gave the closest to the noise shaped reference. But even with the psuedo Gaussian, there was still a noticeable degradation in depth and detail resolution performance compared to the noise shaped reference.

Having a hard time understanding what "noticeable degradation in depth and detail resolution performance" means, just the usual audiophile mumbo jumbo?

 

[bennetng]

Regarding Triangular vs Gaussian dither, as long as the destination bit depth is 24-bit or higher, the noise level differences at analog output should be minimal, so it should not look like this:

The 2/4x plots look like 16-bit quantization to me. Don't know what is happening here but probably some compatibility issues between the M Scaler and third party DACs.

Since my old Adobe Audition 1.5 is capable of dithering down to 4-bit with different PDFs, I attached some audio samples. "Shaped Triangular" in Audition (also called Highpass TPDF in some other software) is same as "Triangular" in Audacity.

This foobar plugin for example offers both regular TPDF and Highpass TPDF dither.

mda Dither DSP (foo_dsp_mdadither) – Component for foobar2000

Download, listen and analyze the attached audio files. All files are converted from "original.wav" to 4-bit. "no dither.wav" is of course glitched, "rectangular.wav" also suffered from noise floor modulation. The other files are mostly fine, the Gaussian ones are slightly noisier, but at 24-bit this difference should not be so dramatic like what this M Scaler review shows. Just to be clear, all the attached files are not noise-shaped, it is another thing.

Also, check out the link in my signature, it is a software capable of detecting and removing unnecessary dither in audio files.

 

[aj625]

There is another measurement/review here. Summary: very poor jitter performance. Filtering-wise; compared to the default Mojo2 filter

CHORD Mojo 2 Review (Portable DAC & HP Amp)

This is a review and detailed measurements of the CHORD Mojo 2 portable, battery operated stereo DAC and headphone amplifier. It is on kind loan from a member and costs US $799. The Mojo 2 is solidly built. While I am definitely not a fan of multi-color interface, the buttons feel better in...

www.audiosciencereview.com

there's no way one could think there is reason for investing on the m-scaler, it won't make the already sharp filter realistically any better.

Amir measurements of filter is at 2x and 4x