• 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: 328 89.9%
  • 2. Not terrible (postman panther)

    Votes: 9 2.5%
  • 3. Fine (happy panther

    Votes: 7 1.9%
  • 4. Great (golfing panther)

    Votes: 21 5.8%

  • Total voters
    365

HP9000

Member
Joined
Aug 20, 2022
Messages
17
Likes
6
This does not mean, that we hear or need 100 kHz. 44.1 kHz sampling rate is perfectly able to reproduce that (and even less than 10 microseconds): https://www.audiosciencereview.com/...s/time-resolution-of-redbook-16-44-pcm.22102/
@Geert "However, it doesn't make any sense to translate this time value into a frequency (the 100 kHz you referred to)"
I distinguished tonal sensitivity from from timing sensitivity, that we do not hear 100khz as a tone but as a harmonic of audible tones. The link you've provided shows differences between left and right channel, which is only half of the problem given that there is still the capability of a given channel to consider.

Fascinating. You have lots and lots of reasons why this device should make an audible difference, and yet it doesn’t.
You say this yet I'll bet you wouldn't buy for example a filter-less NOS dac over a delta sigma dac - because it sounds different.

@voodooless "That's why you need magnitude AND phase, as I mentioned.", "Distortion is distortion.. We can measure it, and it covers all kinds of distortion, also in timing."
The comment of mine you are quoting only mentions how only a frequency-magnitude can be unhelpful for the same reason that providing both a frequency-magnitude and phase plot can still be unhelpful. Take for example an IIR Bessel lowpass filter, a frequency chart will show roll-off, a phase plot will show nonlinear, yet the damping factor/square wave will prove superior to say a Butterworth filter of the same order. You can say that we should be able to infer that it is a Bessel lowpass filter that we're looking at even if all we had were the frequency and phase plots on hand, but that would be tedious in comparison to just being shown the step response.

"So we need oversampling after all?"
I was specific about how higher sample rate music would be provided natively (recorded that way) and not as a matter of oversampling, and if I were to say we need delta-sigma dacs after all, then I would be saying that we need oversampling after all.

"We don't listen to an impulse response. We listen to actual music"
All you're saying is that ringing is not detectible, which I would disagree with. Also, this sounds like when people say "I don't listen to measurements, I listen to music", which usually doesn't come unironically from this site.

"an oversampling DAC can and will reproduce the original waveform better than your NOS DAC, regardless of the sample rate. There is no disadvantage."
A NOS dac provided with natively high sample rate material and having an IIR lowpass filter vs a delta-sigma dac with the same material played into it: the delta-sigma dac's oversampling is redundant and results in higher than necessary ringing, the NOS dac filter's may have non-linear phase, but it's going to have the waveform with less ringing.

"No, the intermodulation is not generated upstream, it's directly output by the NOS DAC."
Hence my use of the word 'supposedly', it is an indictment of what is commonly conceived as the cause because what is upstream to the dac is only tangential.

"If you mean intermodulation products that reflect back to the audible part: obviously they are there, because a NOS cannot properly reconstruct the original waveform."
What do you mean by "reflect back"?

"What? I have no idea what you mean?"
My understanding of intermodulation is that the frequencies that would be labeled such can't count as harmonics. I have heard this worded the same way I put it, "not a part of the original signal". Of course that can be said about any form of distortion, but here it is used for emphasis.

"No, it just shoves the shit up in frequency, making it less audible.", "Sample rate != time resolution. As long as you think it is, you're lost..."
To the first comment: What does a delta-sigma do better than oversample? To the second comment: Yet I am supposed to think that a delta-sigma (read over-sampling) dac has better time resolution than a filter-less NOS dac as a result of something to do with the sample rate.

@Geert "Also, ITD refers to the difference in arrival time of an angled sound source between the 2 ears as a result of different path lengths. It's not a value that indicates absolute time resolution sensitivity, like an impulse being early without a difference in path length."
"An impulse being early without a difference in path length", Path lengths equals time difference. The relevance of there being two ears is that it provides further isolation of the sounds received by either, so that they do not interfere in space. Yet, they are only so isolated from each other before they are going to the same brain. The result is one waveform even if two ears are receiving two different sounds that are isolated before being summed.

@tmtomh "It always strikes me how much incorrect info about digital audio can be traced back to the stair-step fallacy."
If the graphs show stair-steps, why should I think otherwise? If the graph for dynamic range of an amp shows 70db, why should I think otherwise?

It's quite a good restaurant I hear too... although timing is quite important there, as in one must book in advance and there is strict table duration. Numbers are finite and the tables are not modular. The food is always on schedule, the gourmet chef ensures the timing is resolute. One must sample each item on the menu to understand the intricate nuance of the flavours... of course some excess sampling may be required for certain dishes. It really is on a grand scale.


JSmith
Nonsense.
 

tmtomh

Major Contributor
Joined
Aug 14, 2018
Messages
1,589
Likes
4,873
@Geert "However, it doesn't make any sense to translate this time value into a frequency (the 100 kHz you referred to)"
I distinguished tonal sensitivity from from timing sensitivity, that we do not hear 100khz as a tone but as a harmonic of audible tones. The link you've provided shows differences between left and right channel, which is only half of the problem given that there is still the capability of a given channel to consider.


You say this yet I'll bet you wouldn't buy for example a filter-less NOS dac over a delta sigma dac - because it sounds different.

@voodooless "That's why you need magnitude AND phase, as I mentioned.", "Distortion is distortion.. We can measure it, and it covers all kinds of distortion, also in timing."
The comment of mine you are quoting only mentions how only a frequency-magnitude can be unhelpful for the same reason that providing both a frequency-magnitude and phase plot can still be unhelpful. Take for example an IIR Bessel lowpass filter, a frequency chart will show roll-off, a phase plot will show nonlinear, yet the damping factor/square wave will prove superior to say a Butterworth filter of the same order. You can say that we should be able to infer that it is a Bessel lowpass filter that we're looking at even if all we had were the frequency and phase plots on hand, but that would be tedious in comparison to just being shown the step response.

"So we need oversampling after all?"
I was specific about how higher sample rate music would be provided natively (recorded that way) and not as a matter of oversampling, and if I were to say we need delta-sigma dacs after all, then I would be saying that we need oversampling after all.

"We don't listen to an impulse response. We listen to actual music"
All you're saying is that ringing is not detectible, which I would disagree with. Also, this sounds like when people say "I don't listen to measurements, I listen to music", which usually doesn't come unironically from this site.

"an oversampling DAC can and will reproduce the original waveform better than your NOS DAC, regardless of the sample rate. There is no disadvantage."
A NOS dac provided with natively high sample rate material and having an IIR lowpass filter vs a delta-sigma dac with the same material played into it: the delta-sigma dac's oversampling is redundant and results in higher than necessary ringing, the NOS dac filter's may have non-linear phase, but it's going to have the waveform with less ringing.

"No, the intermodulation is not generated upstream, it's directly output by the NOS DAC."
Hence my use of the word 'supposedly', it is an indictment of what is commonly conceived as the cause because what is upstream to the dac is only tangential.

"If you mean intermodulation products that reflect back to the audible part: obviously they are there, because a NOS cannot properly reconstruct the original waveform."
What do you mean by "reflect back"?

"What? I have no idea what you mean?"
My understanding of intermodulation is that the frequencies that would be labeled such can't count as harmonics. I have heard this worded the same way I put it, "not a part of the original signal". Of course that can be said about any form of distortion, but here it is used for emphasis.

"No, it just shoves the shit up in frequency, making it less audible.", "Sample rate != time resolution. As long as you think it is, you're lost..."
To the first comment: What does a delta-sigma do better than oversample? To the second comment: Yet I am supposed to think that a delta-sigma (read over-sampling) dac has better time resolution than a filter-less NOS dac as a result of something to do with the sample rate.

@Geert "Also, ITD refers to the difference in arrival time of an angled sound source between the 2 ears as a result of different path lengths. It's not a value that indicates absolute time resolution sensitivity, like an impulse being early without a difference in path length."
"An impulse being early without a difference in path length", Path lengths equals time difference. The relevance of there being two ears is that it provides further isolation of the sounds received by either, so that they do not interfere in space. Yet, they are only so isolated from each other before they are going to the same brain. The result is one waveform even if two ears are receiving two different sounds that are isolated before being summed.

@tmtomh "It always strikes me how much incorrect info about digital audio can be traced back to the stair-step fallacy."
If the graphs show stair-steps, why should I think otherwise? If the graph for dynamic range of an amp shows 70db, why should I think otherwise?


Nonsense.

1. We don’t hear 100kHz, period. Doesn’t matter if it’s a fundamental or a harmonic.

2. Stairstep graphs are a common, but inaccurate, representation. The sample values shown in such graphs are accurate, but the stair steps drawn to connect them are not. A 70dB SINAD graph is also an abstraction, but it’s an accurate representation.

This is basic stuff.
 

Geert

Major Contributor
Joined
Mar 20, 2020
Messages
1,289
Likes
2,043
I distinguished tonal sensitivity from from timing sensitivity, that we do not hear 100khz as a tone but as a harmonic of audible tones.

An analog 12 kHz sine wave sounds exactly the same as 12 kHz square wave (referring to the step response that started the discussion), so no we don't hear higher than 20 kHz harmonics. No' more of these crazy arguments without references please.

The relevance of there being two ears is that it provides further isolation of the sounds received by either, so that they do not interfere in space. Yet, they are only so isolated from each other before they are going to the same brain. The result is one waveform even if two ears are receiving two different sounds that are isolated before being summed.

You missed the point. Theory about spatial location does not apply to absolute temporal resolution where both ears hear the same stimuli without a path length difference (again referring to that step response you asked for, which normally is identical between left and right).
 

HP9000

Member
Joined
Aug 20, 2022
Messages
17
Likes
6
Thanks for providing such a good summary of your rather lengthy post.
Your comment is pithy, but not a rebuttal. Also, I don't see what's wrong with replying to everyone at once.

1. We don’t hear 100kHz, period. Doesn’t matter if it’s a fundamental or a harmonic.
Yet the references @Geert and I provided say that we can.
However, it doesn't make any sense to translate this time value into a frequency (the 100 kHz you referred to) and suggest the Fourier transfer of a square wave is incomplete. It's like saying I had lunch at 0,083 Hz.
A time value like 10us and 100khz mean the exact same thing, the only translation being made are the abbreviations which are synonymous. This is because frequency denotes to an interval/time window, which is just what the list of seconds, milliseconds etc. denote to. Same denotation different connotations, and not to an extent to where the Fourier transfer need be mentioned.

Why would a reference provide us with this number and state it as audible, yet it is not audible in the form of a fundamental or harmonic? What else are we hearing other than a waveform? And I am using the word 'harmonic' in the specific sense that it refers to the rise time of a waveform. A harmonic may not always be the rise time, but rise time is nonetheless a way that harmonics can take place.

And your last sentence is so key - it's what folks seem to persistently forget or neglect when they make these "time resolution" claims. It always strikes me how much incorrect info about digital audio can be traced back to the stair-step fallacy.
2. Stairstep graphs are a common, but inaccurate, representation. The sample values shown in such graphs are accurate, but the stair steps drawn to connect them are not. A 70dB SINAD graph is also an abstraction, but it’s an accurate representation.
My comments regarding this subject so far have not depended on the conception that the sample points are connected by straight lines across, only that there are so many of the points in a second.

Can you explain the bet? I don’t understand what you mean
This is what you said:
Fascinating. You have lots and lots of reasons why this device should make an audible difference, and yet it doesn’t.
It read like you were disagreeing with the idea that a device measuring better in time should be audibly advantageous. Now I'm not sure if that was a disagreement or just a statement.

An analog 12 kHz sine wave sounds exactly the same as 12 kHz square wave (referring to the step response that started the discussion), so no we don't hear higher than 20 kHz harmonics. No' more of these crazy arguments without references please.
This sounds like you're saying that phase shift doesn't matter as much at higher frequencies, which isn't what I disagree with. I am talking more about lower frequencies that aren't necessarily bass frequencies (though bass frequencies included), such as the lower midrange and upper mid range (most sensitive). The importance of a device being able to produce a 12khz square wave like you're saying is because it can be a byproduct of an ability to play very well the frequencies lower than that. In other words, the capability to play a 12khz square wave can't be 'designed out' of the device, and it shouldn't be frowned upon as if the engineer thinks it's important we hear a 12khz square wave.

You missed the point. Theory about spatial location does not apply to absolute temporal resolution where both ears hear the same stimuli without a path length difference (again referring to that step response you asked for, which normally is identical between left and right).
This kind of timing is being researched in studies about auditory temporal resolution or perception. For normal hearing and low frequencies the time resolution is 3 ms at best ('Temporal Resolution in Normal-Hearing and Hearing-Impaired Listeners Using Frequency-Modulated Stimuli', John P. Madden, Lawrence Feth, 1992).
I can find a slightly larger number in terms of temporal resolution, and that is 22.7-25 microseconds (mono, a-spatial). This is still way beyond single digit milliseconds.
 

NTK

Major Contributor
Forum Donor
Joined
Aug 11, 2019
Messages
1,822
Likes
3,789
Location
US East
From Dr. David Griesinger, slide 35

griesinger.png
 

voodooless

Master Contributor
Forum Donor
Joined
Jun 16, 2020
Messages
5,741
Likes
9,432
Location
Netherlands
A time value like 10us and 100khz mean the exact same thing, the only translation being made are the abbreviations which are synonymous. This is because frequency denotes to an interval/time window, which is just what the list of seconds, milliseconds etc. denote to. Same denotation different connotations, and not to an extent to where the Fourier transfer need be mentioned.
1661201470828.gif

That’s it for now, more at another time.
 

Thomas savage

Grand Contributor
The Watchman
Forum Donor
Joined
Feb 24, 2016
Messages
10,224
Likes
16,046
Location
uk, taunton
View attachment 226073
That’s it for now, more at another time.
Careful with that , those 2 got married in the end , in real life .. ol HP90000000000 might get the wrong Idea but then again Maybe some romancing is what this thread needs .
 
Last edited:

Sir Sanders Zingmore

Addicted to Fun and Learning
Forum Donor
Joined
May 20, 2018
Messages
766
Likes
1,434
Location
Melbourne, Australia
It read like you were disagreeing with the idea that a device measuring better in time should be audibly advantageous. Now I'm not sure if that was a disagreement or just a statement.
I was saying that despite all of the reasons you give as to why this device should be audibly advantageous (or even audibly different), it isn’t.
This device makes no audible difference
 

Blumlein 88

Grand Contributor
Forum Donor
Joined
Feb 23, 2016
Messages
16,444
Likes
28,337
Carful with that , those 2 got married in the end , in real life .. ol HP90000000000 might get the wrong Idea but then again Maybe some romancing is what this thread needs .
I thought she became the time travelers wife? ;)
 
OP
amirm

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
39,501
Likes
180,010
Location
Seattle Area
Why would a reference provide us with this number and state it as audible, yet it is not audible in the form of a fundamental or harmonic?
The author is providing 10 microseconds as the timing differential of signal arrival for localization. This has nothing to do with perceiving frequencies corresponding to such timing. You can create a dual tone: 1 khz and 100 kHz. Chop off the 100 kHz and it will sound the same. Indeed much of the music you listen to has been subject to such transformation in the conversion to 44.1 kHz.
 

HP9000

Member
Joined
Aug 20, 2022
Messages
17
Likes
6
View attachment 226073
That’s it for now, more at another time.
Carful with that , those 2 got married in the end , in real life .. ol HP90000000000 might get the wrong Idea but then again Maybe some romancing is what this thread needs .
I thought she became the time travelers wife? ;)
What I was saying is not that controversial. Say we have a line that represents 1 second. If you have a frequency like 10hz it can be displayed as alternating positive and negative half cycles sized equally and so that with their ends lined up 10 pairs of them fit into that line with nothing left over. If you have time windows of 1/10 sec it can be represented by brackets sized equally and so that with their ends lined up 10 of them fit into that same line with nothing left over. There will be the same number of cycles (frequency) as time windows (intervals), and they will occur in the same place.
You can decide that the half cycles be the focus, so that it doesn't matter whether they all be of a positive or negative polarity, but this would be changing our mind on the frequency; half the cycle, double the frequency. Everything in the above paragraph stays true.

You can create a dual tone: 1 khz and 100 kHz. Chop off the 100 kHz and it will sound the same.
Take a 25khz sine wave (no harmonics) half cycle (now 50khz), split it in half (now 100khz) from the line of symmetry (vertical), have those two parts be the rate at which the half cycles for 1khz reach say 70% of their peak and lower from 70% of their peak. The would-be flat top of the implied square wave here can just be comprised of arbitrary frequencies whose peaks make up the remaining 30%. This is more in the vein of what I'm getting at.
Adding a 100khz sine wave to a 1khz sine wave the way you're describing is as if 100khz should mean a high number of repetitions in the shape - that is, reflecting the amplitude change over time - of the 1khz sine. A tonally inaudible high frequency like 100khz needn't repeat more than four times for what I am talking about.

This device makes no audible difference
People have compared filter-less NOS dacs with delta-sigma dacs, reported audible differences, and even stated some advantages. How I imagine one might reply to this on this site is by saying that they tricked themselves or that their brain did, that the distortion gave to a sound effect or that they're enjoying the distortion. The reason I am not quick to dismiss these findings for those reasons is because a double blind test would have to be conducted comparing both. I also find it intriguing that the one area in which a filter-less NOS dac can be said to measure better than delta-sigma (impulse response) aligns with the language those reports use, such as "fast, realistic".
 

Blumlein 88

Grand Contributor
Forum Donor
Joined
Feb 23, 2016
Messages
16,444
Likes
28,337
What I was saying is not that controversial. Say we have a line that represents 1 second. If you have a frequency like 10hz it can be displayed as alternating positive and negative half cycles sized equally and so that with their ends lined up 10 pairs of them fit into that line with nothing left over. If you have time windows of 1/10 sec it can be represented by brackets sized equally and so that with their ends lined up 10 of them fit into that same line with nothing left over. There will be the same number of cycles (frequency) as time windows (intervals), and they will occur in the same place.
You can decide that the half cycles be the focus, so that it doesn't matter whether they all be of a positive or negative polarity, but this would be changing our mind on the frequency; half the cycle, double the frequency. Everything in the above paragraph stays true.


Take a 25khz sine wave (no harmonics) half cycle (now 50khz), split it in half (now 100khz) from the line of symmetry (vertical), have those two parts be the rate at which the half cycles for 1khz reach say 70% of their peak and lower from 70% of their peak. The would-be flat top of the implied square wave here can just be comprised of arbitrary frequencies whose peaks make up the remaining 30%. This is more in the vein of what I'm getting at.
Adding a 100khz sine wave to a 1khz sine wave the way you're describing is as if 100khz should mean a high number of repetitions in the shape - that is, reflecting the amplitude change over time - of the 1khz sine. A tonally inaudible high frequency like 100khz needn't repeat more than four times for what I am talking about.


People have compared filter-less NOS dacs with delta-sigma dacs, reported audible differences, and even stated some advantages. How I imagine one might reply to this on this site is by saying that they tricked themselves or that their brain did, that the distortion gave to a sound effect or that they're enjoying the distortion. The reason I am not quick to dismiss these findings for those reasons is because a double blind test would have to be conducted comparing both. I also find it intriguing that the one area in which a filter-less NOS dac can be said to measure better than delta-sigma (impulse response) aligns with the language those reports use, such as "fast, realistic".
Filter-less NOS Dacs I would expect to sound different in blind testing. No news there. It is a broken by design DAC. In fact the reason they don't sound worse is because we have such attentuated hearing at higher frequencies so our own hearing response functions as something of a brickwall filter.

The few microseconds timing ability of a pair of human ears does not require response to 100 khz or more. Simple as that. You either don't understand the difference or are trying to troll us. Redbook CD can manage much smaller time discrmination and replay accurately also without 100 khz response.
 

HP9000

Member
Joined
Aug 20, 2022
Messages
17
Likes
6
Filter-less NOS Dacs I would expect to sound different in blind testing. No news there. It is a broken by design DAC. In fact the reason they don't sound worse is because we have such attentuated hearing at higher frequencies so our own hearing response functions as something of a brickwall filter.

The few microseconds timing ability of a pair of human ears does not require response to 100 khz or more. Simple as that. You either don't understand the difference or are trying to troll us. Redbook CD can manage much smaller time discrmination and replay accurately also without 100 khz response.
Broken by design how?

I already stated that 100khz isn't necessary but that there is room for improvement.

I am not talking about spatial discernment, someone else brought that up, I am talking about percussive discernment at the very least.
 

Sir Sanders Zingmore

Addicted to Fun and Learning
Forum Donor
Joined
May 20, 2018
Messages
766
Likes
1,434
Location
Melbourne, Australia
People have compared filter-less NOS dacs with delta-sigma dacs, reported audible differences, and even stated some advantages. How I imagine one might reply to this on this site is by saying that they tricked themselves or that their brain did, that the distortion gave to a sound effect or that they're enjoying the distortion. The reason I am not quick to dismiss these findings for those reasons is because a double blind test would have to be conducted comparing both.
the fact that there does not exist a single properly documented double blind test showing an audible difference lends enormous weight to the theory that perceived differences are not caused by the reasons you (and Watts) suggest.
 

HP9000

Member
Joined
Aug 20, 2022
Messages
17
Likes
6
the fact that there does not exist a single properly documented double blind test showing an audible difference lends enormous weight to the theory that perceived differences are not caused by the reasons you (and Watts) suggest.
I think it's still concerning that it seems a test hasn't been conducted. The variables would have to be fixed, the dacs' metrics would have to be akin to each other in all ways but the impulse response, even if that means contriving for higher distortion (by design) in one.
 

Sir Sanders Zingmore

Addicted to Fun and Learning
Forum Donor
Joined
May 20, 2018
Messages
766
Likes
1,434
Location
Melbourne, Australia
I think it's still concerning that it seems a test hasn't been conducted. The variables would have to be fixed, the dacs' metrics would have to be akin to each other in all ways but the impulse response, even if that means contriving for higher distortion (by design) in one.
??
Surely all that’s required is to switch the M-scaler in and out in a properly conducted (level matched etc) DBT
 

solderdude

Grand Contributor
Joined
Jul 21, 2018
Messages
12,687
Likes
28,469
Location
The Neitherlands
Broken by design how?
Because it lacks a proper reconstruction filter.
Of course upsampling (Chord or software) fixes the 'broken by design' parts.

Perfectly fine to listen to a filterless NOS DAC as the Chord (or software) adds the reconstruction filter a filter-less NOS DAC is missing.
 
OP
amirm

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
39,501
Likes
180,010
Location
Seattle Area
Take a 25khz sine wave (no harmonics) half cycle (now 50khz)
No, that waveform will have infinite bandwidth now due to discontinuities at the edges of half cycles. Rest of your paragraph seems like some kind of automated text generation. Do you have a real signal processing background or just making stuff up?
 
Top Bottom