• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many 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: 358 88.2%
  • 2. Not terrible (postman panther)

    Votes: 13 3.2%
  • 3. Fine (happy panther

    Votes: 7 1.7%
  • 4. Great (golfing panther)

    Votes: 28 6.9%

  • Total voters
    406

Jomungur

Member
Joined
Jan 3, 2022
Messages
92
Likes
591
The Dave, TT2 or Qutest would suffice as they all have dual BNC inputs. 765kHz through a Y adapter is not the same.

I almost want to report this for spreading falsehoods. Anyone believing this should look at Chord's own literature, including the manual for the M Scaler and the Hugo2. (See my quoted excerpt in an post above). The Chord Hugo 2 manual also indicates the input sphere will glow light blue when it is in 'coax dual data mode'. What else could this be? My original intention when sending this unit to Amir was to help someone decide whether they want to buy a standalone DAVE vs. the cheaper M Scaler + Hugo 2.

I can find nothing that supports this quote. If the quoted statement is right, shame on Chord for not correcting their manual that says the M Scaler can be used with Hugo 2 up to 768hz.

This is from the Chord Hugo 2 spec sheet:

Connectivity (input): Micro USB (White): 44.1kHz to 768kHz – 16bit to 32bit
Coax via 3.5mm jack (Red): 44.1kHz to 384kHz – 16bit to 32bit
Coax via (same) 3.5mm jack (Dual data mode: Yellow): 44.1kHz to 768kHz – 16bit to 32bit
Optical (Green): 44.1kHz to 96kHz – 16bit to 24bit


I hope someone at Chord at least reads this thread and least fixes their manual. On the website it says dual data mode yellow, but in the Hugo 2 manual it says dual data mode is light blue. I think this is the main reason why Chord's cycling color scheme is a bad idea: it's arbitrary and confusing, even to Chord's employees apparently. It makes it difficult to test: is coax dual data mode yellow (per the website) or blue (per the manual)?
 
Last edited:

Purité Audio

Master Contributor
Industry Insider
Barrowmaster
Forum Donor
Joined
Feb 29, 2016
Messages
9,051
Likes
12,150
Location
London
Apparently the manual was written by a kid on ‘work experience’ perhaps he also designed the M-Scaler while he was there.
Keith
 

goat76

Major Contributor
Joined
Jul 21, 2021
Messages
1,269
Likes
1,385
I repeat, in the Chord presentation, there was *definitely* a difference - I thought the music had a deeper and more atmospheric soundstage as RW claims on 'The Wam' site apparently. What stumped me (but none of the other listeners it seems) is that the mean volume seemed slightly reduced with the M-Scaler working. sadly, I wasn't able to return and listen again under my terms and as I said earlier, I've had similar things with absolute phase in systems sensitive to it and slight level differences, especially with half a minute or so between A and B, will screw it up for the listener anyway. As for trusting the ears alone (and the mind behind them) to tell things that measurements can't - FECK OFF - especially if you're over 60 years old (or close to it). Thank heavens in my case as *everything* now 'sounds' wonderful and eff subjective differences now :)
I participated in a small blind test upscaling with HQPlayer, I think we were just five people that did the test. It was 3 songs with 3 different files each, two with upscaling at different levels, and a third without upscaling (if I remember it correctly?).

Anyway, the task was to hear which files of each song had the highest upscale option. Two of the participants nailed all three, I got it right on two of the songs, and the two people who were the most skeptical about the test said they didn't hear any differences at all.

I'm aware of the limitations of that little test, it was mostly done just for fun. Maybe @Belker still got those test files he made about 3 years ago? :)
 

fatoldgit

Active Member
Joined
Feb 29, 2020
Messages
279
Likes
326
So for a dude with a basic level of understanding of why we upsample, we do so for three reasons (correct me please!!!):

1- esp for 44.1, we do so to move the brick wall filtering well away from the 22k "zone" and also to ensure aliasing isnt audible

AND/OR

2- to allow us to bugger around with different filters (change pre/post ringing etc) and then upsample these to the native resolution of the DAC so the DAC wont impart its own filtering

AND/OR

3- we believe that upsampling in itself increases resolution and hence provides for better sound reproduction


Can someone please explain how a million taps is better than some lessor number (yes I can google that but I will see marketing gobbledygook that might not provide for an honest appraisal).

Thanks,

Peter
 

Blumlein 88

Grand Contributor
Forum Donor
Joined
Feb 23, 2016
Messages
20,523
Likes
37,054
So for a dude with a basic level of understanding of why we upsample, we do so for three reasons (correct me please!!!):

1- esp for 44.1, we do so to move the brick wall filtering well away from the 22k "zone" and also to ensure aliasing isnt audible

AND/OR

2- to allow us to bugger around with different filters (change pre/post ringing etc) and then upsample these to the native resolution of the DAC so the DAC wont impart its own filtering

AND/OR

3- we believe that upsampling in itself increases resolution and hence provides for better sound reproduction


Can someone please explain how a million taps is better than some lessor number (yes I can google that but I will see marketing gobbledygook that might not provide for an honest appraisal.

Thanks,

Peter
3 is dead wrong. No additional resolution or you would be creating information not in the original which you cannot do.

2 is sort of kind of true.

1 is sort of part of 2. Also in a DAC it is imaging. Aliasing is for ADC's. Sorry for being the pedant. If aliasing occured in the original recording neither upsampling or anything else will remove it. You can upsample to prevent imaging during playback and to reduce phase shifts of being so close to the cutoff of the filter.


PS: ADC aliasing is when sound above 20 khz gets mixed in as a lower frequency below 20 khz. DAC imaging is when below 20 khz sounds get mirror imaged above 20 khz. This for 44.1 and 48 khz sample rates. Cut-off moves proportionally higher for higher sample rates.
 
Last edited:

mansr

Major Contributor
Joined
Oct 5, 2018
Messages
4,685
Likes
10,700
Location
Hampshire
Ignored is that many ADCs use half band filtering, expecting another half band upon playback to get ADC/DAC combined filtering steep enough to get rid of aliasing and imaging. If you upsample and playback then you actually are missing some out of band attenuation you would otherwise get during playback.
Sorry, but that doesn't make sense. The amount of aliasing is entirely dependent on the ADC filter (you even said so yourself), and imaging is entirely determined by the interpolation filter.
 

Blumlein 88

Grand Contributor
Forum Donor
Joined
Feb 23, 2016
Messages
20,523
Likes
37,054
Sorry, but that doesn't make sense. The amount of aliasing is entirely dependent on the ADC filter (you even said so yourself), and imaging is entirely determined by the interpolation filter.
Okay, I'll change it.
 
OP
amirm

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,368
Likes
234,386
Location
Seattle Area
Okay, I'll change it.
What do you mean you are going to change it? For a change I wanted to see an argument not involving me in a review thread!!! :D
 

Blumlein 88

Grand Contributor
Forum Donor
Joined
Feb 23, 2016
Messages
20,523
Likes
37,054
What do you mean you are going to change it? For a change I wanted to see an argument not involving me in a review thread!!! :D
I changed the post and removed the part Mansr pointed out as wrong.
 
Last edited:

JSmith

Master Contributor
Joined
Feb 8, 2021
Messages
5,153
Likes
13,214
Location
Algol Perseus
This whole thread is making me feel aliased... also a bit tapped out. :cool:
Can someone please explain how a million taps is better than some lessor number
That's the point here, it's not "better" and completely unnecessary as once the limits of audibility are reached and what an amplifier is capable of, there is simply no need to go any higher. Well unless once can hear differences at -300dB... :facepalm:


JSmith
 

srkbear

Major Contributor
Forum Donor
Joined
Nov 16, 2021
Messages
1,029
Likes
1,445
Location
Dallas, TX
Beat to death in:


:)
Well, sort of, yes, perhaps no—but great! That’s my Cliff’s Notes summary of Amir’s analysis in that piece (which was expertly done). It doesn’t really address the particular fallacy in dispute here, and it largely concluded that the extant data on high resolution audio is comprised of a bunch of poorly-designed studies on both sides, which basically means no data. I finally embraced his concluding point that higher resolution formats bring us better quality masters and left it there.

Either way, Nyquist calculations aside It seems that we’d all have to gulp hard to dismiss any claims of benefit above 44/16 sampling when there’s scarcely been a DAC reviewed on this site in quite some time that doesn’t resolve sampling frequencies as high as 384 :)
 

Blumlein 88

Grand Contributor
Forum Donor
Joined
Feb 23, 2016
Messages
20,523
Likes
37,054
Did a simple real test of upsampling, and started another thread. Hopefully we didn't scare off @fatoldgit so perhaps this might make things more clear or provoke some questions.

 

Belker

Member
Joined
Feb 20, 2021
Messages
42
Likes
103
Location
Sweden
I participated in a small blind test upscaling with HQPlayer, I think we were just five people that did the test. It was 3 songs with 3 different files each, two with upscaling at different levels, and a third without upscaling (if I remember it correctly?).

Anyway, the task was to hear which files of each song had the highest upscale option. Two of the participants nailed all three, I got it right on two of the songs, and the two people who were the most skeptical about the test said they didn't hear any differences at all.

I'm aware of the limitations of that little test, it was mostly done just for fun. Maybe @Belker still got those test files he made about 3 years ago? :)
Wow, I totally forgot about that test! No I don’t have the files, I think. They were very big DSD files. But anybody can do this. Just download the trial version of HQPlayer Professional. It lets you do off-line conversion with the same features as the desktop “consumer” HQPlayer. The only limitation to the trial is that it shuts off after 30 minutes, then you need to restart it. At least that is how it was when I did this.
 

charleski

Major Contributor
Joined
Dec 15, 2019
Messages
1,098
Likes
2,239
Location
Manchester UK
So for a dude with a basic level of understanding of why we upsample, we do so for three reasons (correct me please!!!):

1- esp for 44.1, we do so to move the brick wall filtering well away from the 22k "zone" and also to ensure aliasing isnt audible

AND/OR

2- to allow us to bugger around with different filters (change pre/post ringing etc) and then upsample these to the native resolution of the DAC so the DAC wont impart its own filtering

Peter
@Blumlein 88 partially addressed this, but I think it's worth clarifying things fully.
It's important to realise that there's no way to avoid the need for filtering at the '22k zone' if your input file is 44.1kHz. You must use a digital filter at that frequency to prevent the production of spurious hf images. Without oversampling you are dependent on the filter that's baked into your DAC, which may or may not be properly designed.

Oversampling allows you to override the 22kHz filter in your DAC and substitute it with one that's better designed (sharper cutoff, better attenuation, linear phase, etc). (The DAC will still be doing some filtering at the oversampled rate, but since this is shifted well into ultrasonics it doesn't really matter.) This is all you're doing with oversampling: replacing the DAC's native filter with a different one. Obviously the resolution of the signal will be unchanged as that's completely dependent on the source material.
 

rkt31

Member
Joined
Jul 22, 2020
Messages
60
Likes
37
@Blumlein 88 partially addressed this, but I think it's worth clarifying things fully.
It's important to realise that there's no way to avoid the need for filtering at the '22k zone' if your input file is 44.1kHz. You must use a digital filter at that frequency to prevent the production of spurious hf images. Without oversampling you are dependent on the filter that's baked into your DAC, which may or may not be properly designed.

Oversampling allows you to override the 22kHz filter in your DAC and substitute it with one that's better designed (sharper cutoff, better attenuation, linear phase, etc). (The DAC will still be doing some filtering at the oversampled rate, but since this is shifted well into ultrasonics it doesn't really matter.) This is all you're doing with oversampling: replacing the DAC's native filter with a different one. Obviously the resolution of the signal will be unchanged as that's completely dependent on the source material.
exactly what m scaler does and it does it better than most upsamplers, as indicated by slope and attenuation !
 

Jomungur

Member
Joined
Jan 3, 2022
Messages
92
Likes
591
I've ordered a Topping D90SE DAC and get it on Saturday. I want to compare it side by side with the DAVE now after more research in my audio systems since it's less than 1/10 of the price. This is just subjective listening tests for myself, but it shows how much I am getting disturbed by Chord's marketing claims.

I revisited the official Chord white paper, which really reads more like a marketing brochure. I've included some quotes from it, and then below them why they seem problematic. Please correct me if I'm not understanding the concepts correctly, as this stuff is new to me (I'm a lawyer, not an engineer).

1. Thirty-five years after the introduction of the CD, some would think that huge leaps forward would be improbable at this point. The Chord Hugo M Scaler proves without doubt that is not the case. It may in fact be the single greatest advancement in digital sound reproduction ever. The Hugo M Scaler offers the world’s most advanced upscaling technology, taking digital files from any source and transforming them into audio that’s virtually indistinguishable from the original analog performance.

This is impossible, right? Through upscaling or upsampling (not sure what the correct term is so I'll use them interchangeably), you cannot recreate the original analog performance from a digital file that was sampled at 44.1hz/16 bit. That is, the best you can do is remove (through sampling, filters, etc.) audible artifacts that are an inevitable byproduct of the analog to digital conversion.

2. At 44,100 samples per second, there’s a gap between samples—22 microseconds to be exact. The problem is musical timing and transient information also occurs in these gaps and whatever information that exists in the gaps is lost when creating the digital file. DACs can’t recover missing timing information—they simply miss the start of the transient. Blurring of transients is the result. That gets confusing for the ear and the brain. Which means we won’t perceive timbre or soundstage or the pitch of bass instruments properly.

This might be true, but not for the reasons stated? That is, the problem with sampling at 44,100khz is that you introduce aliasing from the folding in of frequencies above (.5 * 44,100khz), which distorts the recreated analog waveform. But the distortion is not created by the gaps between samples, the gaps are mathematically filled in by reconstructing the wave form through the inverse Fourier transform? And you can address the aliasing to a large degree through filters and perhaps upsampling.

3. The Hugo M Scaler acts like a “pre-DAC”. It takes the digital file and repairs it, adding back the information lost between the samples, then it sends the repaired file to the DAC. The M Scaler increases the sampling rate from 44,100 times per second (44.1 KHZ) by a multiple of 16, to 705,600 times per second (706.6 KHz). With 705,600 samples per second, a huge amount of important information that was lost when creating the 44.1 digital file is now recovered. The more samples, the closer you get to the original analog signal. In essence the Hugo M Scaler places 15 additional new musical samples in between each original musical sample resulting in an astounding improvement in the recreation of the original music signal.

This is impossible? You cannot "repair" a digital file and add back information that was never there; what you can do through upscaling is mitigate sonic artifacts by shifting the Nyquist threshold to higher frequencies.

4. Rob Watts, Chord’s Digital Design Consultant, has developed his exclusive WTA (Watts Transient Alignment) technology, which incorporates the most advanced interpolation filter of its kind in the world. That mammoth processing power allows for a huge breakthrough in what’s known as tap length of the filter—to a previously unimaginable 1,015,808 taps.

Perhaps a little unfair because this was written a few years ago, but HQ Player can match or exceed the number of taps listed here? More importantly, this implies that the *way* the interpolation filter works matters more than the simple number of taps, and yet it boasts about the number of taps, not sure which factor is meaningful (or both). That is, can filter "quality" matter vs. filter quantity?

5. Simply put, taps are a measure of device’s capacity to reproduce the original waveform. The longer the tap length of the filter, the closer it gets to the original analog signal.

This is incorrect, right? Tap lengths help to a point but they yield diminishing or no returns on reconstruction of the analog waveform?

6. With this ingenious technology, the M Scaler doesn’t do a crude interpolation like all other filters or “guess” to fill in the dots between each step. It peers deep into the actual data itself, as if looking under a microscope, and reconstructs the missing waveform in its exact original form thus creating an almost perfect new digital version of the original analog performance.

This is simply false, and perhaps absurd?

7. The Hugo M Scaler quite literally represents the realization of a lifelong dream for Rob Watts. Half a million lines of code and hundreds of listening tests later, listeners can now experience something they never could before— a huge difference in resolution, bass definition, sound staging, instrument separation and focus and more varied instrument timbre.

Perhaps (it's a subjective claim), but it seems the hundreds of listening tests where done on one (or a few) persons who have an interest in promoting the device?

8. Works With Everything. The Hugo M Scaler is inserted in your system ahead of the DAC— it is not a DAC and it does not replace your DAC. The M Scaler improves the sound of all digital audio systems. It works with all digital files and streaming services and all digital source components; streamers, smart devices, computers, CD/DVD players and video systems. The M Scaler works with all DAC brands so you don’t need a Chord Electronics DAC to get the benefits of upscaling. Even if you have an older DAC that only accommodates 4 or 8 times upscaling, you will still get a very worthwhile improvement in sound quality when adding the M Scaler to your audio system.

This is just plain false, based on testing but also Rob's own posts in the Headfi forums.

I was going to include the testimonials in the white paper but decided not to because it feels like kicking a dog at this point. They are pretty outrageous though.
 
Last edited:
Top Bottom