Audibility of total harmonic distortion: A test

[Vineethkumar01]

Harmonic distortion is audible in varying percentages depending on the frequency and the loudness at which a frequency is played. So, if we look at an equipment, it’s important to know that at what frequency it’s having a particular percentage of distortion at what level of loudness.

This is exactly the reason why I mentioned kef r11 as the best possible speaker we can buy under 5 lakhs. It literally has no distortion which is audible to anyone in any frequency band.

By default when I auditioned it it didn’t wow me immediately, and ended up choosing another speaker. kef has a treble roll off and the speaker immediately found dull in comparison to anything I compared it with in a store. Then I posted about this on another platform, and someone pointed this distortion figure to me and although it was too late reluctantly I went back to the store and this time when streaming from my phone to a dac their i used the eq to push the highs and I realized what I overlooked, it outperformed the speaker I bought and even with the eq there is no grain or any audible flaw of any kind.

If anyone is running a room eq, it’s very important still to have a very low distortion as sometimes the eq applied would push certain frequencies to very higher levels and distortion rises with rise in loudness in most cases, on higher frequencies for many speakers this window is very low.

I learned this the hard way after making a very expensive (to me) mistake. The only reason I made the other post which started a conversation , (sorry @Vineethkumar01 for making a harsh comment there) was I don’t want another person to make the same mistake. I cannot the speaker name here (as It might defame the brand and hurt /trigger guys who own it)but it’s something stereophile reviewed this year and mentioned a bargain at around 7k dollars apparently he didn’t had the distortion graph, and to the horror I found it from a paid test which I bought later after my stupid decision of buying

Hi,
I have no issues regarding you finding that the KEF R11 speakers are to your liking. I too like many speakers, some just by their looks
I also don't doubt your good intentions regarding helping out people by advising them to make informed purchase decisions.
I criticized you in that post just because you were making conclusions about certain objective parameters without looking at enough data. Ultimately being objective means one would need to make conclusions based on relevant measurements right. In that post I could find only a single in-room measurement of that speaker made at 1m along with a THD plot.
For that speaker, we should have at least looked at this much: https://www.soundstagenetwork.com/i...ents-kef-r11-loudspeakers&catid=77&Itemid=153

It could be a good speaker for home use. But to objectively conclude that, we need more data. Because that speaker is meant to be listened to in the far field (all those driver responses should integrate well at listening position), a single measurement at 1m is not of much use. Its sensitivity is not that high compared to what we usually call as high sensitivity speakers. We need to see the directivity index (DI), Early reflections DI (ERDI) and power response to assess its smoothness and make useful comments about its performance at a distance, let's say 2.5m or more. For assessing timing related things, we need step response, impulse response, and excess group delay measurements to say something valid. To conclude something about presence/absence of internal resonances (since it is a big speaker), we need to look at cumulative spectral decay and/or impedance measurements. To make useful conclusions about its EQ ability, we need to look at directivity index smoothness, power compression (like what Erin gives), and distortion measurements. It would be better if we have intermodulation distortion measurements also as it is supposedly more audible than THD. And so on. Amir says his set of measurements are enough. But experienced speaker designers have repeatedly pointed out that he needs to make conclusions with at least more data and even then it could be incomplete.. In fact the whole reason why he (even though half heartedly) has started giving time domain measurements is because of Kimmosto's fights..

Thanks
Vineeth

 

[Sean de Silva]

Hi all,
I found this distortion test on ASR recently. I found it very interesting:

It basically helps us understand at what level of THD each of us, through our audio systems, are able to audibly make out total harmonic distortion.
My personal score as per this video (on my cheap 600Rs edifier earphones) is as follows:

Music: around 11% THD
Tone at 500 Hz: around 5%
Tone at 100 Hz: around 11%

Please test and post your results..

Thanks for sharing this test.
My score
Music 7.913% 500Hz 1.396% 100Hz 0.805%
Is my hearing intact?

 

[mbhangui]

Thanks for sharing this test.
My score
Music 7.913% 500Hz 1.396% 100Hz 0.805%
Is my hearing intact?

as per your figures, you and me are in the same boat, but I feel my ears cannot make out the muddiness for something below 50 Hz. I like listening to music with my subwoofer and my cutoff frequency set my AMP and also allo piano 2.1 is 60 Hz. So my subwoofer gets only frequencies 60 Hz and below. But whenever my subwoofer is turned on, my son detects it immediately and says the overall sound becomes muddy and horrible. This is something I'm not able to detect. How I figured out this problem was a post shared by @Analogous where the speaker says that not everyone detects the muddiness in bass.

A very interesting podcast on Subwoofers

I learnt so much from this fascinating discussion. Have a listen and see for yourself

www.hifivision.com

 

[superczar]

For once, I will disagree with the premise of this supposedly objective test.

Unlike real distortion (e.g. say overdriven Amp), The primary waveform here is playing undistorted on your speakers.
So if you are playing the tracking at an average of 70db, added distortion (which is the rough equivalent of raising the noise floor) is only going to start appearing discernible at 4-5db or more (i.e. 5-7%).. maybe 1-2 db if you have especially sensitive ears but surely not under that.

An amp that is distorting on the other hand would end up affecting the primary waveform itself - and thus would become noticeable at a much smaller threshold.. probably at as low as 0.1 or max 0.5% .

 

[Vineethkumar01]

For once, I will disagree with the premise of this supposedly objective test.

Unlike real distortion (e.g. say overdriven Amp), The primary waveform here is playing undistorted on your speakers.
So if you are playing the tracking at an average of 70db, added distortion (which is the rough equivalent of raising the noise floor) is only going to start appearing discernible at 4-5db or more (i.e. 5-7%).. maybe 1-2 db if you have especially sensitive ears but surely not under that.

An amp that is distorting on the other hand would end up affecting the primary waveform itself - and thus would become noticeable at a much smaller threshold.. probably at as low as 0.1 or max 0.5% .

Hi,
I don't have much knowledge about amplifier related anomalies. So I want to understand more and hence the following questions:

1) When you say real distortion, what does it mean?
To my knowledge when signal clipping occurs at some stage in the signal chain, odd harmonic distortion occurs. For a single tone signal (a sinewave of a particular frequency), looking at the time domain signal, we can identify this by the clipped peaks of the sinewave. If we look at the spectrum of this signal, the distortion will show up as an infinite number of odd harmonic components of the fundamental tone.

2) To my understanding we are trying to understand our ability to test our audible thresholds for total harmonic distortion in this test. So in the video, they make us hear a single tone, and then slowly introduce harmonic distortion by some means. The more the amount of harmonic distortion, the more will be the amplitude of those peaks around the original single peak. At whatever level of distortion we are able to pick up the difference in the sound compared to the sound of the single tone, that corresponds to the minimum distortion we are able to identify audibly. This is the purpose of the test to my knowledge. The distortion can come from any component in the signal chain. So whether we are able to identify it or not is being tested here.

3) In all cases, if we look at the plot, we can see that the noise floor remains below -120 dB. Before and after adding distortion. Only the harmonic distortion components' amplitudes are increased. So I didn't understand how we can say that adding distortion means raising noise floor. Yes, raising noise floor is a form of distortion. But it is not 'harmonic' (harmonic components means integer multiples of the fundamental) distortion to my understanding.

Thanks
Vineeth

 

[superczar]

Hi,
I don't have much knowledge about amplifier related anomalies. So I want to understand more and hence the following questions:

1) When you say real distortion, what does it mean?
To my knowledge when signal clipping occurs at some stage in the signal chain, odd harmonic distortion occurs. For a single tone signal (a sinewave of a particular frequency), looking at the time domain signal, we can identify this by the clipped peaks of the sinewave. If we look at the spectrum of this signal, the distortion will show up as an infinite number of odd harmonic components of the fundamental tone.

2) To my understanding we are trying to understand our ability to test our audible thresholds for total harmonic distortion in this test. So in the video, they make us hear a single tone, and then slowly introduce harmonic distortion by some means. The more the amount of harmonic distortion, the more will be the amplitude of those peaks around the original single peak. At whatever level of distortion we are able to pick up the difference in the sound compared to the sound of the single tone, that corresponds to the minimum distortion we are able to identify audibly. This is the purpose of the test to my knowledge. The distortion can come from any component in the signal chain. So whether we are able to identify it or not is being tested here.

3) In all cases, if we look at the plot, we can see that the noise floor remains below -120 dB. Before and after adding distortion. Only the harmonic distortion components' amplitudes are increased. So I didn't understand how we can say that adding distortion means raising noise floor. Yes, raising noise floor is a form of distortion. But it is not 'harmonic' (harmonic components means integer multiples of the fundamental) distortion to my understanding.

Thanks
Vineeth

I think you have probably misconstrued my post.
This specifically is the reason I qualified my post by calling it the rough equivalent of adding a noise floor.

The THD rating of an amplifier for a given power level is simply a meta / derived metric to measure how well the amplifier fares in amplifying a signal without distorting it.
And while its the harmonic distortion that is being used as a measure to provide a quantifiable difference against a comparable baseline ( so that a user can take an informed call )- it does not mean that an amp with a THD rating of 10% at 80db is adding 10% worth of Only harmonic distortions.


Realistically speaking, we often see long raging threads here about the subtle variations introduced by say speaker cable A vs B .. or well, even power cable C vs D
While I won’t dwelve into that, all I would say is that it would take some really serious hearing chops and an immense amount of concentration to distinguish between two reasonably good speaker cables, if at all they can be distinguished

However practically every single member here (or even a casual listener) would be able to almost instantly distinguish between music playing from a amp with a rated THD of 10% at 78db vs another thats rated a reasonable 0.1% or 0.01% - without even needing to focus at all.
(78db because i presume most would have tested around this level)

Now contrast that with this test where even serious listeners are just about able to detect a difference for the music tracks at levels as high as 7-10%
That tells you something about the design of the test itself, doesn’t it?

 

[Vineethkumar01]

I think you have probably misconstrued my post.
This specifically is the reason I qualified my post by calling it the rough equivalent of adding a noise floor.

The THD rating of an amplifier for a given power level is simply a meta / derived metric to measure how well the amplifier fares in amplifying a signal without distorting it.
And while its the harmonic distortion that is being used as a measure to provide a quantifiable difference against a comparable baseline ( so that a user can take an informed call )- it does not mean that an amp with a THD rating of 10% at 80db is adding 10% worth of Only harmonic distortions.


Realistically speaking, we often see long raging threads here about the subtle variations introduced by say speaker cable A vs B .. or well, even power cable C vs D
While I won’t dwelve into that, all I would say is that it would take some really serious hearing chops and an immense amount of concentration to distinguish between two reasonably good speaker cables, if at all they can be distinguished

However practically every single member here (or even a casual listener) would be able to almost instantly distinguish between music playing from a amp with a rated THD of 10% at 78db vs another thats rated a reasonable 0.1% or 0.01% - without even needing to focus at all.
(78db because i presume most would have tested around this level)

Now contrast that with this test where even serious listeners are just about able to detect a difference for the music tracks at levels as high as 7-10%
That tells you something about the design of the test itself, doesn’t it?

I think it is clearer to me now about what you tried to say initially. That an amp or some other component generating high THD may not only be generating THD but may also be doing something else badly that it could become audibly bad. This could be true. I am not sure because I don't have much experience comparing these aspects much. All I know for now is that there are experts who have said that harmonic distortion alone might not be much audibly offensive as long as it is reasonably low (w.r.t the frequency we consider). All I know about amps and their specs is what is written here (and some some subsequent and preceding posts): https://www.hifivision.com/threads/...-tonal-balance-and-dynamics.87736/post-986562

"However practically every single member here (or even a casual listener) would be able to almost instantly distinguish between music playing from a amp with a rated THD of 10% at 78db vs another thats rated a reasonable 0.1% or 0.01% - without even needing to focus at all.
(78db because i presume most would have tested around this level)"

Regarding this, I wont be so sure given that there are tube amps etc which often have bad rep among some people who say that they generate high harmonic distortion. Yet some people find it very enjoyable.

If anything the use of that video that I posted is educational. It helps one understand what is harmonic given a plot showing it. It also enables one to identify, audibly, when harmonic distortion reaches certain levels that he/she is able distinguish distortion from signal. Given that even many objectivists claim things about harmonic distortion without ever knowing what it is, I find the video very useful

 

[superczar]

I think it is clearer to me now about what you tried to say initially. That an amp or some other component generating high THD may not only be generating THD but may also be doing something else badly that it could become audibly bad. This could be true. I am not sure because I don't have much experience comparing these aspects much. All I know for now is that there are experts who have said that harmonic distortion alone might not be much audibly offensive as long as it is reasonably low (w.r.t the frequency we consider). All I know about amps and their specs is what is written here (and some some subsequent and preceding posts): https://www.hifivision.com/threads/...-tonal-balance-and-dynamics.87736/post-986562

"However practically every single member here (or even a casual listener) would be able to almost instantly distinguish between music playing from a amp with a rated THD of 10% at 78db vs another thats rated a reasonable 0.1% or 0.01% - without even needing to focus at all.
(78db because i presume most would have tested around this level)"

Regarding this, I wont be so sure given that there are tube amps etc which often have bad rep among some people who say that they generate high harmonic distortion. Yet some people find it very enjoyable.

If anything the use of that video that I posted is educational. It helps one understand what is harmonic given a plot showing it. It also enables one to identify, audibly, when harmonic distortion reaches certain levels that he/she is able distinguish distortion from signal. Given that even many objectivists claim things about harmonic distortion without ever knowing what it is, I find the video very useful

Regarding your point on tube amps, very true indeed.
They add harmonic distortions at a far greater amplitude/level than other forms of distortion thus making the sound more pleasing , at least to
some. And at a minimum, will not sound bad/grating to the rest of the folks either.

On the other hand, An AB or a D measured at 5% THD @ (say) 75dB will sound quite horrible to most people (not just audiophiles )

My beef with the video is that it seems deliberately misleading - many viewers may run the test and walk away feeling that the THD rating on ALL amps is inconsequential and thus not a factor worth considering for their next purchase/upgrade.

And AB/D are what comprise 99% of amp sales

 

[Vineethkumar01]

Regarding your point on tube amps, very true indeed.
They add harmonic distortions at a far greater amplitude/level than other forms of distortion thus making the sound more pleasing , at least to
some. And at a minimum, will not sound bad/grating to the rest of the folks either.

On the other hand, An AB or a D measured at 5% THD @ (say) 75dB will sound quite horrible to most people (not just audiophiles )

My beef with the video is that it seems deliberately misleading - many viewers may run the test and walk away feeling that the THD rating on ALL amps is inconsequential and thus not a factor worth considering for their next purchase/upgrade.

And AB/D are what comprise 99% of amp sales

I don't understand why we presume here that many people are going to be mislead by this video. Most people are smart enough to make their own decisions and judgements.
People who go by subjective aspects dont care about what these numbers say.
People who go by objective aspects are supposed to know what these numbers mean.
People who take a bit of both find this interesting to learn about.

In the video, it is never mentioned that one should buy things based on this test. So i guess we should take it for what it is and not over interpret things.

 

[raghupb]

I assumed the objective of the tests were to ascertain how much our ear-brain can forgive rather than what the equipment can or cannot do.
Cheers,
Raghu

 

[superczar]

I don't understand why we presume here that many people are going to be mislead by this video. Most people are smart enough to make their own decisions and judgements.
People who go by subjective aspects dont care about what these numbers say.
People who go by objective aspects are supposed to know what these numbers mean.
People who take a bit of both find this interesting to learn about.

In the video, it is never mentioned that one should buy things based on this test. So i guess we should take it for what it is and not over interpret things.

Reason to call it deliberately misleading is right there in the first thirty seconds of the video.. or even in the title .

Right at the beginning, , he calls it an “Audio distortion test” (not harmonic distortion)
then he sets the premise by saying he will play music and test tones with “Increasing amount of distortion” till you can see when you (viewer) can start to hear the “distortion”

But said distortion is added by a synthetic inclusion of harmonic distortion alone - Something that is very easy to miss unless one is paying close attention to the test methodology

 

[alpha1]

The choice of music in this test is very poor.
Lets hear the same test on clean acoustic instruments ...