• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required as is 20 years of participation in forums (not all true). There are daily reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

CHORD M-Scaler Review (Upsampler)

Rate this product:

  • 1. Poor (headless panther)

    Votes: 318 90.3%
  • 2. Not terrible (postman panther)

    Votes: 8 2.3%
  • 3. Fine (happy panther

    Votes: 7 2.0%
  • 4. Great (golfing panther)

    Votes: 19 5.4%

  • Total voters
    352

PassionforSound

Member
Reviewer
Joined
Jul 24, 2022
Messages
45
Likes
11
The two measurements I've flagged don't prove that and yet it's already being judged.
Not quite. The M-scaler is being judged negatively because the M-Scaler is objectively meaningless.
 

KSTR

Major Contributor
Joined
Sep 6, 2018
Messages
2,190
Likes
4,809
Location
Berlin, Germany
Ultimately, the problem is what has been said many times in this thread: the M Scaler is impossible to really test in terms of measurements because it's impossible to quantify what is it is supposed to do. Even the transients point doesn't seem to hold up in measurements.
I do not think this is true. Adding to @dc655321's post IMHO it is totally clear what the idea is: provide the best possible approximation to the ideal, the true sinc() reconstruction via upsampling, basically replacing the downstream DAC's filter. In a hardware blackbox, besides the cheaper but less convenient option of doing this software.

True sinc() reconstruction takes the DAC fully out of the picture (except effect of its analog post-filter), what comes out are the sample values the recording ADC generated, plus exact waveform in between the samples. Therefore, the only filter function in place is that of the ADC, a combination of the analog pre-filter and the internal filter used for downsampling to the target output rate.
[Sidenote: Again, this process can be offloaded to an extra step by running the ADC oversampled and then apply a perfect sinc() filter (in software) for downsampling]

Obviously, this point becomes moot if the music is highly processed and/or non-acoustic source. However, for well-recorded acoustic music and the like, this device fulfills the whole idea of high fidelity, do not harm the original signal.

One practical drawback of true sinc() filtering/resampling is that you you need the highest amount of headroom for intersample-overs which is why the digital output level of the device is lowered.

As for audibility of all this, after a few tests I did in software (with a special version of the SoX tool) I'd say no, sorry.
 
Last edited:

the_brunx

Active Member
Joined
Apr 7, 2020
Messages
268
Likes
653
I tried to warn you guys, this is a man on a mission, he’s not here for enlightenment or truth, the truth about snake oil in audio is the very thing that make he’s snake oil promoting YouTube channel a hypocrite. He realeases well scripted videos almost every 2-4 days. This is probably he’s only Job. How will he then ever accept the truth? Waste of time, I will now sit back and watch where this all goes
 

mansr

Major Contributor
Joined
Oct 5, 2018
Messages
4,601
Likes
10,330
Location
Hampshire
IMHO it is totally clear what the idea is: provide the best possible approximation to the ideal, the true sinc() reconstruction via upsampling
What constitutes best, though? Is it stopband attenuation? Passband flatness? Transition sharpness? Since we can't maximise all of those at the same time, what compromises are acceptable? Rob Watts takes a very simplistic "more taps better" view, and like the sheep in Animal Farm, his acolytes are happily bleating this mantra. It is, however, anything but clear that his approach has any merits compared to traditional methods.
 

Geert

Major Contributor
Joined
Mar 20, 2020
Messages
1,167
Likes
1,819
The two measurements I've flagged don't prove that and yet it's already being judged.

I already explained you the scope of this judgement, being noise and distortion performance. And as explained, all parameters being measured are judged. Also when they're only relevant from a technical perspective. Hence the evaluation of jitter performance. So why insist?
 
Last edited:

Jomungur

Member
Joined
Jan 3, 2022
Messages
91
Likes
569
I do not think this is true. Adding to @dc655321's post IMHO it is totally clear what the idea is: provide the best possible approximation to the ideal, the true sinc() reconstruction via upsampling, basically replacing the downstream DAC's filter. In a hardware blackbox, besides the cheaper but less convenient option of doing this software.

True sinc() reconstruction takes the DAC fully out of the picture (except effect of its analog post-filter), what comes out are the sample values the recording ADC generated, plus exact waveform in between the samples. Therefore, the only filter function in place is that of the ADC, a combination of the analog pre-filter and the internal filter used for downsampling to the target output rate.
[Sidenote: Again, this process can be offloaded to an extra step by running the ADC oversampled and then apply a perfect sinc() filter (in software) for downsampling]

Obviously, this point becomes moot if the music is highly processed and/or non-acoustic source. However, for well-recorded acoustic music and the like, this device fulfills the whole idea of high fidelity, do not harm the original signal.

One practical drawback of true sinc() filtering/resampling is that you you need the highest amount of headroom for intersample-overs which is why the digital output level of the device is lowered.

As for audibility of all this, after a few tests I did in software (with a special version of the SoX tool) I'd say no, sorry.
I see, thanks. Well, I'm not a sound engineer so I am probably way ahead of my skis in trying to understand this. I can tell you this, though, most dealers have no idea what it does.

It's also paradoxical. From what you say, the higher the resolution of the source file, the better an upscaler should work because it has more to work with. It can't make gold out of lead, all it can do is polish all the dust off the gold that it is presented with. And yet the higher the resolution of the source file, the less the need for upsampling, right, because the file needs less "help"? That goes to the point many have made about it solving a non-problem, I suppose.
 
Last edited:

Jimi Floyd

Member
Joined
May 5, 2022
Messages
58
Likes
177
Location
Pisa, Italy
the M Scaler is impossible to really test in terms of measurements because it's impossible to quantify what is it is supposed to do. Even the transients point doesn't seem to hold up in measurements.
no, no, no. Please don't mind if I correct some errors contained in your post. you should have written a correct summary of the evidence: "It is possible to test the M-scaler in terms of measurements. Some of these tests have shown objective defects, such as increased jitter on digital outputs and induced distortion in a non-Chord DAC connected to the M-scaler. Other standard measurements of digital devices performed on the M-scaler showed no benefit to upscaling." I find this phrasing more correct as a thread digest, thank me later.
 
OP
amirm

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
39,111
Likes
174,558
Location
Seattle Area
- the jitter measurement which you stated would be fixed by any decent DAC
This product has been marketed as perfecting the perfection of their DACs. Now we hear that it doesn't have a great PLL and spits out jitter?

Why would this be an argument for me to not measure jitter? Was I to assume it is great or terrible? I measure because I can't take for granted what a company advertises. And good thing too in this case.

There are a ton of audiophiles that would run out of a room if they thought a device had "jitter." Just look at any streamer review where the more expensive the streamer, the less jitter they say it produces. No way what you stated is accepted by the camp that promotes these expensive products.
 
OP
amirm

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
39,111
Likes
174,558
Location
Seattle Area
- the 1kHz tone which you stated did not benefit from upsampling
We only know this after I tested it. Prior to that, you wouldn't know if noise floor is way elevated for example. Or modulated as I discovered in the DAVE DAC yesterday. It could have also overloaded and clipped with 0 dBFS signal. And again, remember that the DAC is now running at higher sample rate and so that would change performance as well.

As I keep explaining, the ideal upsampler does no harm. Real ones may indeed do so. As such, it is critical to test for that. And testing I did and found out the performance of Topping D70s was degraded.

I still don't know why you are advocating that we should know less, than more.
 
OP
amirm

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
39,111
Likes
174,558
Location
Seattle Area
Let's pretend for a moment that the M-Scaler does do what it's said to. If someone new to the hobby we're looking around for reviews and saw your comments on the 1kHz tone test (with Hugo 2) and on the jitter test, they'd likely be instantly turned away because they wouldn't necessarily understand that neither measurement matters.
Instantly? Why wouldn't they read the full review? BTW, they should be turned away because the entire measurement set shows that the M-scaler can actually degrade performance. And at best, it doesn't do anything other than provide a sharper reconstruction filter.

I suspect the people you worry about won't understand any of that so I provided audio samples. Said sample was with music so there should be no excuse.
 
OP
amirm

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
39,111
Likes
174,558
Location
Seattle Area
So, I guess my question here is this. Are you looking to provide objective data and unbiased interpretation of that data or are you using the data to illustrate your own opinions of the products? (I am not judging either approach and not suggesting any manipulation of the data)
My opinion has nothing to do with it. Or I would not bother testing any of these tweaks.

When I see unique products like this, I devise a series of tests:

1. Do no harm tests. Such tweaks better not in any form or fashion make things worse. I routinely find expensive audio cables that let in more noise for example than cheap one. And that comes out of such tests.

2. Specific tests for the function of the device. Here, you see that in filter tests where we found issues with high noise floor. Company confirmed this online and in your interview. That they used a noisier gaussian dither. Rob claims this is good for sound but he doesn't have a shred of evidence to provide. And literature shows that he is mistaken in this.

You talk about bias. What was yours going into this interview? Why not pause the interview the moment he said he hasn't read my measurements? Let me read them and then speak intelligently about them. You two made no mention of my listening tests and music samples provided. Only snide remarks that I don't listen. That was wrong and showed extreme bias, right?

Net, net, I go into every one of these tests with a blank sheet of paper. If there is a difference, I find it. And differences I found, albeit, not favorable to this product. I also confirmed and gave credit for what it does do (sharper filtering). You keep pointing out the extra tests I ran and ignore the ones that were specific to it being an upsampler. That other tests are not complimentary to the product is not my issue. I rather be complete and comprehensive in my testing.

Finally, let me remind you again that company has provided exactly zero measurements of its own. If you wanted me to verify their claims, then they should have provided such.
 
OP
amirm

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
39,111
Likes
174,558
Location
Seattle Area
Finally, I want to clarify here that I reached out to Rob Watts to ask him to answer my questions about the measurements, not to read and critique your review.
I don't see the distinction whatsoever. You asked him questions about my review and measurements and even showed the snapshots. This was the entire scope of the long interview. You also set it up as reading my review and thinking the product doesn't perform. I can't believe you are walking back from this.
 

iamsms

Member
Joined
Dec 14, 2020
Messages
31
Likes
113
That they used a noisier gaussian dither. Rob claims this is good for sound but he doesn't have a shred of evidence to provide. And literature shows that he is mistaken in this.
This is the part I don't understand - RW claims this on head-fi and maybe other places, then goes on to do an interview (or maybe before), and nobody questions him "what is your source for this claim". Not even the interviewer who is supposed to play the role of a journalist here.

A general issue I have with audiophiles and more importantly reviewers - make a claim that sounds plausible (Gaussian dither gives better soundstage, class A sounds fuller etc. etc.) - associating subjective impressions with technologies we engineers studied/worked on for years, and then provide absolutely no reference, no mathematical proof or even do the bare minimum of controlled blind testing.

Either go for total subjective and give your impression and refrain from making unproven association (gaussian dither -> better soundstage, class A -> better dynamics) or go for objectivity. Otherwise this is nothing more than clorox/horse dewormer for recent pandemic all over again.

Sorry for the rant.
 

Geert

Major Contributor
Joined
Mar 20, 2020
Messages
1,167
Likes
1,819
Please stop assuming you understand my motivations or have any real sense of my competency or knowledge.

If you have the knowledge then I seriously wonder why this review is your main concern, and why you were not more critical for all the questionable statements Watts made? Upsampling having an effect on timing and depth? Small signals needing at least -301dB accuracy? Unlimited sensitivity for noise floor modulation? No published reseach or blind tests?

Cynically the recent review of the Chord Dave DAC shows it probably has the worst noise floor modulation ever, next to some other deficiencies. Curious how you're going to tackle that.
 

PassionforSound

Member
Reviewer
Joined
Jul 24, 2022
Messages
45
Likes
11
I already explained you the scope of this judgement, being noise and distortion performance. And as explained, all parameters being measured are judged. Also when they're only relevant from a technical perspective. Hence the evaluation of jitter performance. So why insist?

I ask this because the content is put forward, as I understand it, as objective and unbiased. The conclusions don't seem to support this and I find that to be misleading so I am seeking clarification as to the intention - nothing more
 

Ken1951

Senior Member
Joined
Sep 28, 2020
Messages
459
Likes
883
Location
Blacksburg, VA
I ask this because the content is put forward, as I understand it, as objective and unbiased. The conclusions don't seem to support this and I find that to be misleading so I am seeking clarification as to the intention - nothing more
Obviously the replies you've gotten don't fit your chosen narrative. They're not going to change. Either accept technical reality, or be elsewhere I think.
 
Top Bottom