Causes of "harsh" sound in speakers (and tweeter material)

[valerianf]

@thanossapiens , could you post a MP3 of the music that triggered some harsh sound?
If it is during high level sound spikes it could be distorsion from one of your equipments that goes to saturation.

 

[thanossapiens]

@thanossapiens , could you post a MP3 of the music that triggered some harsh sound?
If it is during high level sound spikes it could be distorsion from one of your equipments that goes to saturation.

Off the top of my head, the parts in these timestamps. especially if its louder than a certain amount

 

[ctrl]

From what I have seen I notice uncomfortable sounds or sibilance if you will, when what plays looks like the image below; something thats much louder than the rest, in mid and tweeter frequencies. Maybe most common around crossovers but I dont think its that because it happens with headphones too. Could also be that these kind of dynamics are too much for the mics in the recording and thats what we are hearing, but I have seen huge variance of how speakers handle this kind of thing so I wonder what the best in this regard are. I'm not sure how to even call this kind of distortion sadly, but it seems to me that soft dome speakers cope better with it.

According to your spectrum screenshot, the harsh sound is not sibilants, which are more likely to be assigned to the 4000-8000Hz region (it might be overtones).

The region around 2.2kHz shown in your screenshot are typical problems of a loudspeaker with errors in radiation behavior and-or too aggressively tuned crossover (on-axis SPL too high in the 2-4kHz range). The on-axis FR can be flat or even have a dip in that area and yet the in-room response can show a hump due to poor speaker radiation in that frequency range.

In my experience, compression or distortion are not the cause if the harsh sound occurs at normal SPL - e.g. at the listening position <80dB@stereo@3m in a normal room, this would correspond roughly to <80dB@1m per speaker, so compression or distortion do usually not play a role at these SPL.

If you want to get to the bottom of it, you would have to search for "moving mic rew room youtube" and record your in-room response for each speaker at the listening position.
If you don't notice any abnormalities (evenly sloping averaged in-room curve without humps) in the 1.5-8kHz range (perhaps there is a problem with the overtones that are really in the sibilant range, so check up to 8kHz), then there is another cause.

 

[valerianf]

I have tried the second You Tube music on 3 systems:
An Android tablet (class D amp).
A desktop computer (class D amp) using Focal Speakers and a subwoofer.
A home theater set up (class AB) using large tower speakers.

I did not ear any harshness or distortion.
For sure the tablet experience was the worst without any low mid-range or bass.

On this link you can see measurements of the Yamaha ns-1000
https://audiokarma.org/forums/index.php?threads/yamaha-ns-1000-x-measurements.310465/
And, surprise: there is a pick of distortion around 2.5 khz.

Good move would be to substitute another type of speaker in your system or strongly suppress the 2.5khz range with an EQ filter.

 

[ctrl]

On this link you can see measurements of the Yamaha ns-1000
https://audiokarma.org/forums/index.php?threads/yamaha-ns-1000-x-measurements.310465/
And, surprise: there is a pick of distortion around 2.5 khz.

Assuming the midrange measurement is correct, it is physically impossible to hear the distortion **.

The fundamental tone was around 93dB during the measurement. The measurement should show HD2 and HD3. HD3 would normally be below HD2.
So we have around 0.2% HD2 and in the worst case 0.2%HD3 at 93dB sound pressure level. In case of 0.2% HD2, it would be easily masked even with a single tone masker.

In case of 0.2% (-53dB) HD3 it is easily masked too. Even in the worst case by a single tone masker. The blue dot in the diagram would be HD3 at -50dB for a 1kHz masker tone at 90dB. The masker tone would be in our case for example a single synthesizer key stroke at 90dB which causes about 40dB HD3 (an additional tone at 3kHz) by playing through the driver which will be completely masked. The masking profile would be comparable for a 2.5kHz masker tone and an additional tone at 7.5kHz (HD3) with -53dB - more detail about masking with links to more details can be found here.

Source: Zwicker, Fastl - Psychoacoustics

So if harmonic distortion do not increase extremely under different radiation angles, it's absolutely inaudible (measured HD2 and HD3)**.

** Since these statements about masking always refer to mean values for a group of test subjects, it is theoretically possible that some individuals may be able to perceive the test tones below the shown threshold (but in no way as "harsh sound")

 

[valerianf]

It is just practical physics that when someone push the sound level or the amplifier distords because reaching it's power limit or it is the speaker distorsion that starts to be too high.
And the first speaker distorsions to be heard is the weakest ones: the lower frequencies and the pics.
It is of no use for negating reality.

 

[ernestcarl]

Honestly I should had phrased my OP post better, since its also not only a tweeter thing.
From what I have seen I notice uncomfortable sounds or sibilance if you will, when what plays looks like the image below; something thats much louder than the rest, in mid and tweeter frequencies. Maybe most common around crossovers but I dont think its that because it happens with headphones too. Could also be that these kind of dynamics are too much for the mics in the recording and thats what we are hearing, but I have seen huge variance of how speakers handle this kind of thing so I wonder what the best in this regard are. I'm not sure how to even call this kind of distortion sadly, but it seems to me that soft dome speakers cope better with it.
You have probably noticed it at its worst if you play something like piano from phone or laptop speakers loudly... that kind of thing.


View attachment 325735

Is it *clipping?

Can you give a music track with time-range example? Assuming it's a flaw in the recording mix/mastering itself, well, you can't do much about it -- but I would create a second copy and EQ the album/tracks if I'm really invested. If it's harshness around the mid-HF xo, then I would apply enough of an EQ cut without making the speakers sound too overly dull in that limited band only.

 

[ernestcarl]

Off the top of my head, the parts in these timestamps. especially if its louder than a certain amount

Ah, just read this... I'll have a listen

 

[ernestcarl]

Off the top of my head, the parts in these timestamps. especially if its louder than a certain amount

Sorry, first link is "not available" in my region so I skipped it. The second music track doesn't sound harsh at moderate listening volumes and is just fine to me. Above 83dB (c-weighted) the those peaks are starting to get just a tad bit annoying -- however, I would attribute this characteristic to the recording itself.

If you want to get to the bottom of it, you would have to search for "moving mic rew room youtube" and record your in-room response for each speaker at the listening position.
If you don't notice any abnormalities (evenly sloping averaged in-room curve without humps) in the 1.5-8kHz range (perhaps there is a problem with the overtones that are really in the sibilant range, so check up to 8kHz), then there is another cause.

An example of moving mic measurements (MMM) taken at different MLPs of the exact same speaker in-room vs single-point sweeps:



Directivity of this coaxial speaker should be flattish on- and off-axis around and well above the mid-HF crossover region. Another perspective with @pierre's generated graphs from the manufacturer's EASE files (10m distance) and default factory settings:

Measurements for speaker Fulcrum Acoustic RX699

www.spinorama.org

And here's yet another perspective using their own spec sheets:

 

[thanossapiens]

Sorry, first link is "not available" in my region so I skipped it. The second music track doesn't sound harsh at moderate listening volumes and is just fine to me. Above 83dB (c-weighted) the those peaks are starting to get just a tad bit annoying -- however, I would attribute this characteristic to the recording itself.

Maybe most speakers are more susceptible to distortion from peaks like that, and especially in that region.
Also I feel like it can suddenly get sibilant etc, i'm not sure if this is clipping or something else. For both of these songs for example in a lot of system and headphones, I can be increasing the volume ever so slightly, like 0.2db each step, and after one it suddenly starts distorting a lot, like from 10% to 100% if that makes any sense