ASR Audio Science Review forum YouTube Channel

[Miska]

12 minutes ago, asdf1000 said:

 And that includes above 20 kHz and up to 100 kHz?

 

Directly or indirectly yes, for example TIM that is very well known to be audible since 70's, works exactly on through those high frequencies.

 

At that time we were seeing exactly same kind of SINAD school as again today. History repeating itself. Amplifier manufacturers were saying that their amps very absolutely perfect because they had very good SINAD. And yet people were hearing clear distortion on transients (broken transients). It took couple of Finnish guys to uncover this and develop a measurement for it. (and later an amplifier that didn't have a problem, based on which actually Electrocompaniet got started)

 

To measure TIM based on the original paper, you need signal that has at least 100 kHz or more bandwidth.

 

[Miska]

6 minutes ago, pkane2001 said:

I assume that's because THD is mostly masked by the signal, while IMD can be far removed from it?

 

Yes, that's the case. While aliasing or IMD is related to the signal in a different way, at frequencies below the fundamental.

 

6 minutes ago, pkane2001 said:

I'll see if I can make up some music samples with IMD content from ultrasonic frequencies. This shouldn't be hard.

 

Let's say you have a difference tone at 2 kHz, it doesn't matter if it's difference of fundamental and it's image around 22.05 kHz or around some higher frequency. But typically it would be more a sequence of harmonics. Best examples you can see by looking at some audio band measurements of NOS DACs operating at 44.1k sampling rate.

 

Does Distort have an aliasing test? It is usually nicely audible by listening to a frequency sweep, you can hear the alias running around.

 

[pkane2001]

4 minutes ago, Miska said:

 

Yes, that's the case. While aliasing or IMD is related to the signal in a different way, at frequencies below the fundamental.

 

 

Let's say you have a difference tone at 2 kHz, it doesn't matter if it's difference of fundamental and it's image around 22.05 kHz or around some higher frequency. But typically it would be more a sequence of harmonics. Best examples you can see by looking at some audio band measurements of NOS DACs operating at 44.1k sampling rate.

 

Does Distort have an aliasing test? It is usually nicely audible by listening to a frequency sweep, you can hear the alias running around.

 

 

DISTORT simulates the non-linearity that produces IMD. I'll just need to create a signal with some different (realistic) levels of ultrasonic signals and with music in the audible range, and then pass it through the non-linearity. I can also disable the anti-alias filter which will allow for aliases in the audible range. These are two different tests, though.

[PeterSt]

23 minutes ago, Miska said:

I have found through "blind test" that any errors or distortions at -120 dB level, and possibly below are audible in audiophile use.

 

We can't believe you because we believe in AmirM. Sorry.

 

[Miska]

7 minutes ago, pkane2001 said:

DISTORT simulates the non-linearity that produces IMD. I'll just need to create a signal with some different (realistic) levels of ultrasonic signals and with music in the audible range, and then pass it through the non-linearity. I can also disable the anti-alias filter which will allow for aliases in the audible range. These are two different tests, though.

 

Both produce sub-fundamental tones in a different way. IMD of images vs aliases. But Bob Stuart also claims in relation to MQA that aliasing is not audible (because MQA has aliasing). Same way as the IMD.

 

There are two cases of IMD, one is from regular straightforward tones, but these are from high frequencies are much lower level. Then another is IMDs between repeating image pairs which is higher level source signal and has different frequency/harmonic system.

 

Either way, multi-tone and sweep/multi-sweep signals are good way for listening to these.

 

[pkane2001]

Just now, Miska said:

 

Both produce sub-fundamental tones in a different way. IMD of images vs aliases. But Bob Stuart also claims in relation to MQA that aliasing is not audible. Same way as the IMD.

 

There are two cases of IMD, one is from regular straightforward tones, but these are from high frequencies are much lower level. Then another is IMDs between repeating image pairs which is higher level source signal and has different frequency/harmonic system.

 

Either way, multi-tone and sweep/multi-sweep signals are good way for listening to these.

 

 

I assume that with multi-tones and sweeps (and white noise?) detection should be easier, with music may be a little harder. 

[PeterSt]

1 hour ago, pkane2001 said:

I grew up around a musical family, Peter. I play the piano, my father was a trained opera singer, my mother an accomplished pianist. My wife is a singer with a wonderful voice.

 

All right Paul. Then I must have misinterpreted the conversation with Frank. And beyond, because I seem to have no problems in hearing anything which is not measurable (that I can do with my equipment and which should be sufficiently good for the job).

 

It won't help, but my background is close to 100% the same (I have the grand of my mother still here). But no singer in the family. Instead I once drove an opera singer from Holland to Poland - she practicing half of the 20 hour trip in the car. Needed more damping. :-)

Both my mother and my father (viola and conservatorium teacher on that) always practicing.

 

Anyway, from such experiences one learns to know what sounds right / realistic, and what does not so or not at all. I never have the need to throw in measurements to know, or worse, have to discuss with e.g. you that what we can't measure, won't not there.

So what is it that grows in my mind (making up differences) which stays out of yours ? something is not right here. I have told it more often (even in this very thread): you are sufficiently eager to the desire of hearing (it). So why your balance seems to be towards the objectivist side ...

no clue (yet).

Superfluously: I measure the h*ell out of myself, spend a lot of money on that, but there is no way that I will ever think that my hearing is off. Finding ways to measure ... that's part of this hobby. And again ... you do that too ...

 

Never mind, 0c.

[Miska]

1 minute ago, pkane2001 said:

I assume that with multi-tones and sweeps (and white noise?) detection should be easier, with music may be a little harder. 

 

Me too. I don't have a dB figure for test signals. The -120 dB figure I have is from music listening.

 

For some other purposes I also did quite a bit of masking tests, for how much below noise levels one can still hear discrete tones and at which frequencies. This is interesting and more related to the passive sonar things I used to work on.

 

For example with -6 dBFS white noise, -26 dB 1 kHz sine is still audible. On a waveform display such of course looks just like noise, but on long enough FFT it is clearly visible too. And on passive sonar spectral display we employ all kinds of noise filtering that can dig out discrete tones deeply buried in noise. The question is always how good these algorithms are vs human operator listening (and how trained the operator is).

 

[asdf1000]

2 minutes ago, Miska said:

human operator listening (and how trained the operator is).

 

Out of interest, have you seen/witnessed these (highly trained) operators hear things at abnormally low levels?

 

That software misses?

 

 

[lucretius]

2 hours ago, pkane2001 said:

Search for "sent to me by the company" without quotes in Amir's posts. You'll need to ignore some where the words are not in the same sentence, but at least six pages of matches. You'll see high-end DACs, headphones, speakers, etc. Obviously, there may be other wording sequences in some of the reviews that this doesn't match.

Emphasis added.

 

I did the search and after limiting it to AMIRM, the 6 pages of matches turns into 3.  From that I identified 27 companies.  I also identified 2 companies not on this list from my earlier review bringing the total to 29 companies that have submitted items to Amir for review.

 

A very large proportion of the equipment submitted for review were DACs costing $200 USD or less.  Only four companies submitted equipment costing more than $1,000 USD, which included:

• 4 DACs (2 from the same company) for $1,149, $1,299, $1,400, and $3,409
• 1 active monitor for $2,200

 

With all due respect, this appears to be a far cry from your statement, "You'll see high-end DACs, headphones, speakers".

 

[asdf1000]

@Miska did you have the title & authors of the TIM paper you mentioned, so I could search.

[Miska]

15 minutes ago, asdf1000 said:

@Miska did you have the title & authors of the TIM paper you mentioned, so I could search.

 

Eero Leinonen, Matti Otala and John Curl: "A Method for Measuring Transient Intermodulation Distortion (TIM)"

 

They have two methods and I use the high quality requirement signal with -3 dB at 100 kHz.

 

Quote from conclusions part:

Quote

many amplifiers having excellent total harmonic
distortion and SMPTE-IM data show high values of
distortion as measured with the proposed method

 

 

Remember that signals with images are stair-stepped at input to the following analog stages.

 

[lucretius]

52 minutes ago, PeterSt said:

I measure the h*ell out of myself

 

Please do not publish this and speak of it no more. 🙂

[Miska]

44 minutes ago, asdf1000 said:

Out of interest, have you seen/witnessed these (highly trained) operators hear things at abnormally low levels?

 

That software misses?

 

Yes, usually those are short term transients. With FFT and such you can dig a lot into noise, but with the cost that the time window becomes very long too. It works well for stationary signals, but not for short term signals much shorter than the time window.

 

There was also some published studies somewhat related to the topic:

https://phys.org/news/2013-02-human-fourier-uncertainty-principle.html

 

[pkane2001]

37 minutes ago, lucretius said:

Emphasis added.

 

I did the search and after limiting it to AMIRM, the 6 pages of matches turns into 3.  From that I identified 27 companies.  I also identified 2 companies not on this list from my earlier review bringing the total to 29 companies that have submitted items to Amir for review.

 

A very large proportion of the equipment submitted for review were DACs costing $200 USD or less.  Only four companies submitted equipment costing more than $1,000 USD, which included:

• 4 DACs (2 from the same company) for $1,149, $1,299, $1,400, and $3,409
• 1 active monitor for $2,200

With all due respect, this appears to be a far car from your statement, "You'll see high-end DACs, headphones, speakers".

 

I'm not sure what you're arguing. My original statement, as I recall was about two dozen manufacturers that sent in their equipment for review. You disagree, and say that actually about 29 companies submitted their items for review. OK. You're right! It's not two dozen, it's more. I could help you with more keywords and search strings if you really want to find additional items, but I'm too lazy and not interested. Really not sure why you are. 

 

 

[lucretius]

20 minutes ago, pkane2001 said:

 

I'm not sure what you're arguing. My original statement, as I recall was about two dozen manufacturers that sent in their equipment for review. You disagree, and say that actually about 29 companies submitted their items for review. OK. You're right! It's not two dozen, it's more. I could help you with more keywords and search strings if you really want to find additional items, but I'm too lazy and not interested. Really not sure why you are.

 

This discussion evolved from my question, "What high-end manufacturer is going to submit their gear for measurement?".  Sorry if that was not apparent. Although I had thought it was clear as you stated in one of your responses, "You'll see high-end DACs, headphones, speakers, etc.".

 

Clearly, manufacturers are not submitting high-end equipment to Amir for review.  Further, precious little mid-range equipment is submitted by manufacturers to Amir for review.

[pkane2001]

3 minutes ago, lucretius said:

This discussion evolved from my question, "What high-end manufacturer is going to submit their gear for measurement?".  Sorry if that was not apparent.  Clearly, manufacturers are not submitting high-end equipment to Amir for review.  Further, precious little mid-range equipment is submitted by manufacturers to Amir for review.

 

Clearly your definition of "high-end" and mine differ. If you're talking about $10K+ DACs and $5K+ headphones, then I agree with you. 

[lucretius]

1 minute ago, pkane2001 said:

 

Clearly your definition of "high-end" and mine differ. If you're talking about $10K+ DACs and $5K+ headphones, then I agree with you. 

 

OK. But even this last statement of yours is somewhat misleading -- the limits are set too high.  Only 2 pieces of equipment exceeding $1,400 were submitted by manufacturers to Amir for review.

• 1 DAC for $3,409
• 1 active monitor for $2,200
• 0 headphones

Again, to be clear, I'm talking about any equipment exceeding $1,500 USD (yes there are 2 outliers), not just $10K+ DACs and $5K+ headphones.

[PeterSt]

1 hour ago, lucretius said:

Clearly, manufacturers are not submitting high-end equipment to Amir for review.

 

Of course not. For him all under 115dB sounds the same anyway, so why let that measure. To capture possible faults ?

 

I showed the Schiit thing at (IIRC) 115uV of distortion. Strange that this happened because someone thought it sounded odd. Oddly vibrant. I forgot the output level of the Egg but I recall that the distortion measured 0.00410% (THD+N - uhh no ... wait ... something with S). 

[pkane2001]

2 hours ago, lucretius said:

 

OK. But even this last statement of yours is somewhat misleading -- the limits are set too high.  Only 2 pieces of equipment exceeding $1,400 were submitted by manufacturers to Amir for review.

• 1 DAC for $3,409
• 1 active monitor for $2,200
• 0 headphones

Again, to be clear, I'm talking about any equipment exceeding $1,500 USD (yes there are 2 outliers), not just $10K+ DACs and $5K+ headphones.

 

Sorry again, I didn't know at what threshold something becomes "high-end". To me, it's high-end if it's more than I would want to pay for it ;)

[semente]

My DAC is malfunctioning. Should I send it to ASR for testing? 🤔

[pkane2001]

17 minutes ago, semente said:

My DAC is malfunctioning. Should I send it to ASR for testing? 🤔

 

If it sounds too good to be true, then by all means 🤪

[Summit]

17 hours ago, pkane2001 said:

You’re missing the point. It’s not that it’s not possible to not hear the replay chain. It’s that you have no idea what the recording is supposed to sound like with a fully transparent system, because it is a hybrid composite of many processing steps, from multi-mic digitization to compression to spatialization to frequency and phase EQ and whatever other things the mastering engineer decides to apply to produce the sound they like or the sound the artist requested.

 

If you think you know what the original recording should sound like — you’re wrong, unless you were there when the recording was produced.

 

This argument comes up often, and no you don’t need to be there when the recording was produced to know if by replacing one piece in a system makes it sound more lifelike and real or not.

 

When evaluating audio you should of course use many different recordings that you are familiar with how the sound in other quality systems. The more references the better.

 

This not limited to audio. The same is true for TV. It is not necessary to have been there when the recording the nature program for example to see if one TV has a more lifelike picture than another, when it comes to accurate skin tone, blackness level, color depth and so on.  

[The Computer Audiophile]

1 minute ago, Summit said:

 

This argument comes up often, and no you don’t need to be there when the recording was produced to know if by replacing one piece in a system makes it sound more lifelike and real or not.

 

When evaluating audio you should of course use many different recordings that you are familiar with how the sound other quality systems. The more references the better.

 

This not limited to audio. The same is true for TV. It is not necessary to have been there when the recording the nature program for example to see if one TV has a more lifelike picture than another, when it comes to accurate skin tone, blackness level, color depth and so on.  

 

I definitely hear what you're saying, but the issue is that we have no clue what it's supposed to sound like. This may be vastly different from what you believe it sounds like in a "real" environment. If we are to reproduce audio as close to the original recording as possible, then we can't really rely on what we think something sounds like in real life unless we were at the recording and we know what was done to it for the final release. 

 

For video, many creators saturate the colors purposely. Our TVs should reproduce that saturation as released by the creators, not try to turn the saturation down in order to reproduce a more realistic picture. 

[Summit]

6 hours ago, pkane2001 said:

 

No, never :) I grew up around a musical family, Peter. I play the piano, my father was a trained opera singer, my mother an accomplished pianist. My wife is a singer with a wonderful voice. I live around live music and go to concerts and have live performances at my house every chance I get. Measurements don't rule, and I never said that. Measurements are the necessary basis for proper sound reproduction because sound is Physics, as is electronics. Without measurements (or a proper reference) you're guessing, and your guess, while possibly completely different, is no better than mine. That was my point to Frank, and the same to you.

 

 

Say what!