Audibility of low frequency distortion in speakers

[Kvalsvoll]

It is interesting that it doesn't ruin the perception of the music, but it certainly isn't accurate and it is easily discernible from the clean version.

Is it "distructive"? Well not per se, but as you allude to the final effect and perception will depend somewhat on the material and the system.

30% and equal level up to 5h is very high distortion, even in the bass range, measurement-wise we would say that we have exceeded the capacity limit of the system.

On pure sinewave detection level for hd is similar to the situation at higher frequencies, but no instrument produces pure sine waves. It can be generated, and some odd productions may have such signals - anyone? examples?

Say we reduce hd to 5%. Still quite high, but now it may be very difficult to hear it. The larger bass-systems can have around 1% thd down to around 20hz, at quite respectable output levels. Distortion from walls and objects vibrating in a room makes it hard to get much lower.

Distortion in itself is not the problem in the bass range. Linear faults in time and frequency domain is by far the most difficult to fix. Capacity IS a real problem, and this also is connected to distortion. When limits are exceeded, transient peaks are lost and you soon get more distortion - >100% - than whatever is left from the original signal.

For full-range speakers, distortion in bass will affect the rest of the frequency spectrum, and modulate midrange frequencies. If bass is removed from the speakers that reproduce midrange, sound quality will be much better, and even if the bass-system distorts quite heavily, sound quality can still be experienced as quite acceptable.

 

[March Audio]

30% and equal level up to 5h is very high distortion, even in the bass range, measurement-wise we would say that we have exceeded the capacity limit of the system.

On pure sinewave detection level for hd is similar to the situation at higher frequencies, but no instrument produces pure sine waves. It can be generated, and some odd productions may have such signals - anyone? examples?

Say we reduce hd to 5%. Still quite high, but now it may be very difficult to hear it. The larger bass-systems can have around 1% thd down to around 20hz, at quite respectable output levels. Distortion from walls and objects vibrating in a room makes it hard to get much lower.

Distortion in itself is not the problem in the bass range. Linear faults in time and frequency domain is by far the most difficult to fix. Capacity IS a real problem, and this also is connected to distortion. When limits are exceeded, transient peaks are lost and you soon get more distortion - >100% - than whatever is left from the original signal.

For full-range speakers, distortion in bass will affect the rest of the frequency spectrum, and modulate midrange frequencies. If bass is removed from the speakers that reproduce midrange, sound quality will be much better, and even if the bass-system distorts quite heavily, sound quality can still be experienced as quite acceptable.

Please refer to the earlier parts of this thread where we have discussed most of this and to understand the premise of the thread.

This is just an investigation to to see how the research literature stacks up with what ASR members can hear. We first looked at sine signals and it certainly does stack up with the research.

30% in music (4th and 5th were not equal to 2nd and 3rd btw) was just a starting figure, an obvious level so that people can get an understanding of the effects of distortion which aren't necessarily as imagined (timbral instead of gross obvious clipping type).

In Amirs speaker tests we have certainly seen speakers hitting around 20% distortion and far, far higher where he insists on running them below port resonance.

So I disagree, distortion is a problem in speakers (one of many). The issue IMO is that people don't always recognise what they are hearing as related to distortion.

Yes the addition of subs can reduce distortion in speakers, but this and other effects are not really what we are discussing.

 

[andreasmaaan]

Ok, in the end it was possible to do the test on the HD6XXs. I suspect it had been undetectable on the earlier attempt due to background noise rather than inherent distortion of the headphones. Late at night, it didn't seem to be difficult.

As discussed earlier the volume doesn't make a significant difference.

Except at very low frequencies (where levels need to be quite high for the distortion components to cross the absolute threshold of audibility), increasing SPL should increase audibility thresholds - this is certainly the case above 40 or 50Hz (and if I understood correctly, this test added distortion up to F1 of 100Hz, right?).

 

[ctrl]

As discussed earlier the volume doesn't make a significant difference.
I think that wether the distortion is "disturbing" somewhat misses the point.
The fact is you could easily discern the difference here from the undistorted version.

It was, for me, only easy to distinguish in the ABX test, since one could quickly switch back and forth between the two versions there.
If I had only listened to version A once and then version B to decide which was the version with distortions, I would have failed for sure.



Here you can find an ABX sample that shows the harmonic distortion profile of the Adam T5V - which was the reason (apart from @amirm's comment on the distortion measurement) for this thread in the first place.

I hope my approach is the same as @March Audio when creating the test sample.
Got the distortion values from Amir's measurement first:

Then I used them in the software Distort and saved the sample:

Directly attached you will find the first sample and in the next post I will attach the distorted sample.

Don't be surprised about the "wrong" signature, I had to shorten the samples again afterwards to be able to upload them.

At low volume, the sample with distortion can be detected well, in direct comparison.

But at 96dB, for example, my result would certainly no longer be so clear, perhaps I would even fail completely. At least this is shown by short listening tests at corresponding sound pressure.
In this example, when listening to music, 15% HD3 is barely audible or not audible at 96dB sound pressure.

Personally I'm looking for accurate reproduction, not timbrally shifted. It doesn't need to be grossly "offensivly"distorted to be "wrong".

In principle, I can fully agree with this, however you have to put the influence of the harmonic distortion in the low bass range in relation to the " timbral shift" of the entire loudspeaker.
Uneven directivity, such as broadening or narrowing in the radiation of the loudspeaker, or an uneven listening window,... certainly has a much greater impact ("timbral shift") than increased harmonic distortion in the low bass range.
Best to avoid all errors
.

 

[ctrl]

Here is the second ABX sample for the listening test of the harmonic distortion of the Adam T5V at 96dB.
.

 

[March Audio]

Here is the second ABX sample for the listening test of the harmonic distortion of the Adam T5V at 96dB.
.

Thanks but for for the moment can we not add additional tests into the thread or start let it wander onto other subjects. This is about low frequency distortion.

 

[March Audio]

Except at very low frequencies (where levels need to be quite high for the distortion components to cross the absolute threshold of audibility), increasing SPL should increase audibility thresholds - this is certainly the case above 40 or 50Hz (and if I understood correctly, this test added distortion up to F1 of 100Hz, right?).

Yes they have to be above the threshold of audibility. Where we have a listening difference between say 85 and 95dB spl, what you might consider a "typical" listening range, the difference isn't significant.

Also remember that the harmonics (2 to 5) stretch up to 500Hz

 

[andreasmaaan]

Yes they have to be above the threshold of audibility. Where we have a listening difference between say 85 and 95dB spl, what you might consider a "typical" listening range, the difference isn't significant.

Close enough for casual experimentation, sure. The real issue will be that harmonics in the 100-500Hz range are going to be significantly less audible at say 85dB than at say 65dB. So if people are doing the test at low-moderate levels (in quiet environments obviously), their results are likely to be more sensitive than if they'd performed the tests at 85dB+.

 

[ctrl]

Thanks but for for the moment can we not add additional tests into the thread or start let it wander onto other subjects. This is about low frequency distortion.

Sorry, but my sample is about the audibility of harmonic distortion in the low frequency range - that's what it's all about, isn't it?
I only took the distortion profile of a specific loudspeaker at 60Hz and 96dB sound pressure.

It should also be legitimate to evaluate the relevance of a topic like "audibility of HD in the low frequency range" in one sentence - compared to other topics for the evaluation of loudspeakers.
Especially since it was a direct answer to you stressing how important it is to pay attention to "accurate reproduction, not timbrally shifted".

Such a short classification also prevents myth-making, especially when the low harmonic distortion of Purifi drivers is on everyone's lips.

I know, when your listening examples will be for "normal" sound pressure level (<85dB) and we then try to distinguish 2% HD3 at 50Hz (which many inexpensive speakers achieve) from the undistorted sample, everyone can make up their own mind.
.

 

[March Audio]

Close enough for casual experimentation, sure. The real issue will be that harmonics in the 100-500Hz range are going to be significantly less audible at say 85dB than at say 65dB. So if people are doing the test at low-moderate levels (in quiet environments obviously), their results are likely to be more sensitive than if they'd performed the tests at 85dB+.

65dB is pretty quiet. I doubt that the majority of people listen to music at that low a level.

We have the data for a more typical listening range as shown earlier in the thread. People are less sensitive at the 100dB level than 80dB.

 

[March Audio]

Sorry, but my sample is about the audibility of harmonic distortion in the low frequency range - that's what it's all about, isn't it?
I only took the distortion profile of a specific loudspeaker at 60Hz and 96dB sound pressure.

It should also be legitimate to evaluate the relevance of a topic like "audibility of HD in the low frequency range" in one sentence - compared to other topics for the evaluation of loudspeakers.
Especially since it was a direct answer to you stressing how important it is to pay attention to "accurate reproduction, not timbrally shifted".

Such a short classification also prevents myth-making, especially when the low harmonic distortion of Purifi drivers is on everyone's lips.

I know, when your listening examples will be for "normal" sound pressure level (<85dB) and we then try to distinguish 2% HD3 at 50Hz (which many inexpensive speakers achieve) from the undistorted sample, everyone can make up their own mind.
.

Im justr asking that we dont ( at the moment) clutter the thread up with a dozen different samples from a dozen different people. Also I was referring to going off piste with other speaker issues/topics.

 

[andreasmaaan]

65dB is pretty quiet. I doubt that the majority of people listen to music at that low a level.

I don't normally listen at that level, at least not with my full attention, but if I'm trying to pass a distortion detection test, I'll tend to adjust the level until the distortion is most audible...

Anyway carry on. As I said before, I think the circumstances here are ok for casual experimentation

 

[March Audio]

Here is another one

https://1drv.ms/u/s!AnQ0c7fb_4zLgVFWUIaGXvGRipZY?e=X1aCwq