CHORD M-Scaler Review (Upsampler)

[theREALdotnet]

Regarding this particular effect (not seen in electronics) I would say tonality (of transducers) is not completely characterized when listening to music by (measured) FR nor phase and furthermore is not properly correctable with EQ either. At least not with 'simple' parametric or graphic EQ.

The shown responses will absolutely be evident from the frequency response (which includes both amplitude and phase response, BTW), but not at the frequency of that tone. They are the result of a combination of high- and low-pass filters with their respective damping characteristics which in total make up the response of the device.

A theoretical pair of headphones that passes this tone burst test with a clean response (no over- or underswings) would be one with a flat frequency response all the way to infinity.

The reason this cannot be EQed away completely in practice (it can in theory) is simply that the device is bandwith limited, and to EQ it into producing frequencies it cannot produce would require infinite energy.

 

[amirm]

What I find particularly odd are the pages of positive Reviews from around the world that sing the M SCALER’s praises. It’s unlikely Rob Watts bribed/baffled them all with technical jargon.

Why not? I can sell a toothpick with a story and get that much positive press. My story just needs to appeal the lay understanding of said reviewers and I am golden.

Have you ever seen an audio product that gets uniformly bad reviews? I haven't either proving my point above.

Wake up and learn how audio works and not be a victim of such marketing.

 

[Lukino]

Why not? I can sell a toothpick with a story and get that much positive press. My story just needs to appeal the lay understanding of said reviewers and I am golden.

Have you ever seen an audio product that gets uniformly bad reviews? I haven't either proving my point above.

Wake up and learn how audio works and not be a victim of such marketing.

If we take it from the other side, Amir is not interested in giving bad recommendations. On the contrary, he would be happy to recommend this device even if it were only a little beneficial in what we hear. Because that's the goal. He doesn't even criticize the price. But if these devices do nothing but consume electricity, it's just the business of these companies that take advantage of people's ignorance and inability to verify things. People like Amir are few and far between, but we need them like salt. There are various reviews in magazines and websites hand in hand with the manufacturer. How will they evaluate the product? Rob Watts should be a world expert, it's interesting that he uses the Dan Clark Stealth metered headphones reviewed by Amir. In this case, is the measurement important? Amir, I wish you good health. We value people like you very much. P.S. Speakers used by RW... 803 bowers not sure if they would pass Amir's test.....

 

[voodooless]

What I find particularly odd are the pages of positive Reviews from around the world that sing the M SCALER’s praises. It’s unlikely Rob Watts bribed/baffled them all with technical jargon.

What a funny guy you are...

Too bad many Forums on the Internet fail to exchange useful information in favor of promoting Herd Mentality.

 

[Lukino]

Reviews....which one would you choose....92 is more than 88 isn't it? Those are good reviews....I can't imagine how I can throw a number while listening to the hearing amplifier....???? Just an example of fig. Topping A90 and Burson Conductor GT... It's probably clear which is which? And which one is better, right?

 

[bennetng]

Reviews....which one would you choose....92 is more than 88 isn't it?

Still much lower than 1,015,808.

 

[solderdude]

The shown responses will absolutely be evident from the frequency response (which includes both amplitude and phase response, BTW), but not at the frequency of that tone. They are the result of a combination of high- and low-pass filters with their respective damping characteristics which in total make up the response of the device.

A theoretical pair of headphones that passes this tone burst test with a clean response (no over- or underswings) would be one with a flat frequency response all the way to infinity.

The reason this cannot be EQed away completely in practice (it can in theory) is simply that the device is bandwith limited, and to EQ it into producing frequencies it cannot produce would require infinite energy.

Characterize yes... but you did notice you mentioned amplitude and phase response so it cannot be fully characterized with just FR.
Explains why I cannot always 100% correlate step and FR plots (there is a relation but not 100%) and the mains reason I post both as well as a very short 'personal impression' of how I perceive the sound which also does not have full relation to my (flawed) measurements, which of course differ from standards and other people's perception.

Yes, BW limited with lots of resonances at various frequencies. In the end it is the actual wavefront arriving at the ears that determines what the brain makes of it.
At least that's what I got from JJ.

I have to admit the math involved goes completely above my head. I see measurement results and correlate.

Of course this all has nothing to do with the Chord which this thread is about. A new thread about perception and measurements would be needed.

 

[theREALdotnet]

Characterize yes... but you did notice you mentioned amplitude and phase response so it cannot be fully characterized with just FR.
Explains why I cannot always 100% correlate step and FR plots (there is a relation but not 100%) and the mains reason I post both as well as a very short 'personal impression' of how I perceive the sound which also does not have full relation to my (flawed) measurements, which of course differ from standards and other people's perception.

Yes, BW limited with lots of resonances at various frequencies. In the end it is the actual wavefront arriving at the ears that determines what the brain makes of it.
At least that's what I got from JJ.

I have to admit the math involved goes completely above my head. I see measurement results and correlate.

Of course this all has nothing to do with the Chord which this thread is about. A new thread about perception and measurements would be needed.

I can’t say I disagree with any of that. And yes, to me, frequency response has always meant amplitude and phase response. That’s how we were taught, and I’m a bit dismayed that the phase response is often omitted where an amplitude response is shown. Both together (in the same diagram) make it much easier to see what you’re dealing with.

And yes, it is right that for certain systems (linear, time-invariant, yadda yadda) the phase information can be derived from the amplitude plot, but you’d have to measure slopes very accurately, which is almost impossible, especially where filters overlap. It‘s just so much easier to have an explicit phase response plot.

 

[bennetng]

Regarding the so-called "illegal" signal, to be practical, file formats like wav and flac neither detect nor reject something like flat-line DC, impulse, white noise and such when users save the files. Therefore downstream equipment are required to deal with these signals. Also, if the audio data is by definition non-bandlimited (e.g. music composed with some types of digital synthesizers), "illegal signal" is really an invalid argument.

In the case of Rob Watts' approach, an overly steep filter generating overly high intersample over in some music genres is a direct consequence of Rob Watts' approach, instead of those music genres are bad because they contain "illegal" signal.

 

[amirm]

That’s how we were taught, and I’m a bit dismayed that the phase response is often omitted where an amplitude response is shown. Both together (in the same diagram) make it much easier to see what you’re dealing with.

There is a difference between textbook and real life. Real life says phase general is immaterial with respect to evaluating audio performance.

 

[solderdude]

Also, if the audio data is by definition non-bandlimited (e.g. music composed with some types of digital synthesizers), "illegal signal" is really an invalid argument.

That would only be the case if music is digitally created and marketed that way without any BW limiting in the mastering process.
For real music recordings all signals (should) go through BW limiting in the ADC process.
Otherwise all aliasing will be embedded in the audio band and there is no way to undo that regardless how 'ideal' the reconstruction is ?

 

[DSJR]

And I’m assuming the vast majority of those on this Forum have never heard DAVE/TT2 with M SCALER to be making a truly informed comment.



I know what I’ hear and what I like and THAT is what matters most to me.

Cheers.

I have and I did (full Chord system with £40k Dynaudio speakers) and came away nonplussed as I felt levels had been changed in th ebefore and after playing of the track. Be very sure that this isn't what you're hearing when you switch it in and out if it works for you, then sit back and enjoy it is all I can recommend. I was grateful for attending the Chord organised dem and a little envious of those able to afford such bling confections without thinking much about it, but it's totally irrelevant to me really.

 

[bennetng]

That would only be the case if music is digitally created and marketed that way without any BW limiting in the mastering process.

Yes. "Mastering" can be as simple as changing level, fade, pan and such to make an album more consistent. These processes do not require bandlimiting and they don't have "less" artistic value. Looks like Rob Watts have no interest in these genres.

[edit] Also, for indie artists or small labels where the role between composers and engineers is rather vague, and when they have full access to every element (track, media clip, plugin, automatable parameter) in a project, things are even more complex. For example, a song can have non-bandlimited synthesized music on the instrument tracks, but bandlimited vocal on other tracks, and different tracks can have different inserts and so on. Vocal with complex background music often has compressor, reverb, EQ and such to bring up the voice of the singer so that the vocal can be clearly heard instead of buried into the music. Music on the other hand depends on arrangement and types of (synthesized) instrument timbre, may not need these treatments, or at least not the same treatment. The master bus also does not always require plugins like limiter and such as long as you are not a loudness warrior.

Therefore the final outcome can be a complex mix of bandlimited and non-limited product.

 

[theREALdotnet]

There is a difference between textbook and real life. Real life says phase general is immaterial with respect to evaluating audio performance.

Sorry, I was thinking a bit more general than audio, but even for audio I’d dispute that. Phase may not matter all that much when it comes to the soundwaves that arrive at our ears (unless you listen in stereo ), but it matters a lot for things like amplifier feedback loops, stability, EPDR, etc.

 

[Snake Oil is life]

 

[theREALdotnet]

View attachment 219693

Taps, schmaps!

 

[Veri]

Taps, schmaps!

View attachment 219695

Only 22k taps? Those are rookie numbers, obviously!

 

[solderdude]

Therefore the final outcome can be a complex mix of bandlimited and non-limited product.

But.. if one does the final conversion to 44.1 (usually records in 48, 96 or 192) doesn't that entire stuff go through a filter in the downsampler ?

Do you have examples ( I figure these would be rare ) of recordings with such illegal steps (assuming it is not a glitch) ?

 

[bennetng]

But.. if one does the final conversion to 44.1 (usually records in 48, 96 or 192) doesn't that entire stuff go through a filter in the downsampler ?

Not all synthesizers supports all sample rates. Even if they support different sample rates, they may sound different when using the same preset (can be interpreted as an "instrument") or when using the same automation data. Automation means changing the sound generator's parameter continuously in time, for example, following the tempo. Automation can also be applied by recording fader / knob input from an external MIDI controller. Therefore even if the synth supports multiple sample rates, changing sample rate may also require laborious retweaking of parameters.

Do you have examples ( I figure these would be rare ) of recordings with such illegal steps (assuming it is not a glitch) ?

Correlation between sample rate and audible frequency?

No -- it isn't ultrasonic. I don't understand why this misconception persists; perhaps some people are confusing noise shaping with ringing. Totally different concepts. The ringing is just an effect of taking out >fs/2 frequencies from the signal. They have no frequency components

www.audiosciencereview.com

 

[Wunderphones]

There is a difference between textbook and real life. Real life says phase general is immaterial with respect to evaluating audio performance.

I had a phase-related question about the M-Scaler, and more generally about Watts's obsession with huge numbers of taps: let's say I've upsampled from CD quality to the 768 level. Now I have 16x the number of "frames" per second (if I may import that term from video), but the frames are all the same. Where does M-Scaler or the downstream DAC get the information required to determine whether a waveform is supposed to start at frame 247 instead of frame 449?