harshness

[Professor smith]

where does the harshness come from in a speaker? what frequency range causes this?

 

[AllenB]

I am not so sure harshness covers just one phenomena. I could generalise and suggest you look at diffraction, or fix sibilance issues. I believe that sibilance issues can be influenced by the response of a wide range of frequencies throughout the midrange and treble. Does your room make things worse?

 

[Professor smith]

yes the room does change the harshness. One room which is more lively but also better dimensionally, gives less harshness. Another room which is bad dimensionally but also much more damped and dead, gives me more harshness. Everything else remaining constant, what does this indicate?

 

[strawberry]

I think what you hear is the harsh peak near the end of the audible range of the driver. You can combat the peak by putting a zobel circuit in parallel with the driver. For a woofer it's typically 8-10 ohm (10 watt) and 14-22 uF 63-100v.

 

[AllenB]

yes the room does change the harshness. One room which is more lively but also better dimensionally, gives less harshness. Another room which is bad dimensionally but also much more damped and dead, gives me more harshness. Everything else remaining constant, what does this indicate?

Not enough to suggest the speaker is free of harshness on its own. Regardless though, room placement might improve lower midrange cancellations whose frequencies fundamentally support many musical sources. Maybe treble reflections are an issue. Maybe the speaker has poor off-axis performance.

I think what you hear is the harsh peak near the end of the audible range of the driver. You can combat the peak by putting a zobel circuit in parallel with the driver. For a woofer it's typically 8-10 ohm (10 watt) and 14-22 uF 63-100v.

I wouldn't suggest this in this way. Zobel compensation is neither a requirement, nor a good idea as a small tweak with an existing non-compensated crossover.

 

[Soundminded]

In my experience the harshness usually comes from a peak in the range of 4 to 5 khz especially if output in that region is even slightly greater than in the region just below it. This translates into a brightness, hardness especially in string instruments (violins in particular), tendency of pianos to sound like harpsicords. A dip at that frequency range translates into a dullness or remoteness. FR in that range is extremely critical. This may be done deliberatly because to untrained ears it is especially attention getting in rapid fire A/B comparisons in dealer showrooms but experienced listeners know it can quickly become very tiresome when listening at home. The best cure is a graphic equalizer. This allows you to find the region where the problem is and correct it. The sharper and narrower the peaks, the more bands you will need so that you can selectively adjust for only those frequencies where the problem is. If your religion precludes using an equalizer, learn to live with it or throw a heavy winter coat over your speakers. That should fix it too.

 

[mdocod]

Amplification quality can have an enormous amount to do with it in my experience. If you are stacking up a pile of odd ordered harmonics, They are apt to culminate in the mid-high frequency range and make the system sound harsh and/or fatiguing (pick your adjective).

If this is not the case, then I agree with the direction others have suggested. Probably dealing with peaking issues where wavelength size is close to that of the diameter of a driver or smaller but before significant cone breakup cancels out the frequency. Best solution IMO is to try to get x-over points low enough to get these peaks at least 6-12dB below the primary listening level. Some drivers tame out these peaks better than others. Also, as I understand, these "phase plugs" in the center of many modern designs are, I believe, intended to help diffuse the on-axis peaks, with varying degrees of success from driver to driver I also like the effects of a zobel network in nearly ever sim attempt I have made, so considering how cheap that is to try I say it's a no brainer.

Eric

 

[gedlee]

where does the harshness come from in a speaker? what frequency range causes this?

Until there is a common deffinition - in objective terms - you cannot define where it comes from. I don't know of any accepted deffinition. I think that we would all agree that it a higher frequency effect, but beyond that I don;t think that there is much consistancy.

 

[Professor smith]

Until there is a common deffinition - in objective terms - you cannot define where it comes from. I don't know of any accepted deffinition. I think that we would all agree that it a higher frequency effect, but beyond that I don;t think that there is much consistancy.

As a science person I agree we need to be objective but we are talking about sound qualities here. How can we expect consistency? What you say, applies to every aspect of sound quality.

 

[bobodioulasso]

I experience harshness from my Audax hm170Z18 (c/o 2200hz) when it's rising response is not properly equalized down.
That's very noticable when listening to piano. It sounds like a resonance around 1000/1500hz.

 

[whgeiger]

SWAG

where does the harshness come from in a speaker? what frequency range causes this?

It comes from overshoot and ringing of an under-damped tweeter, or if low frequency information is not sufficiently suppressed in the drive signal.

Regards,

WHG

 

[gedlee]

As a science person I agree we need to be objective but we are talking about sound qualities here. How can we expect consistency? What you say, applies to every aspect of sound quality.

In the world of psychoacoustics there are many terms that have well defined meanings. Brightness, Roughness, etc. are all well defined, but harshness is not IMO.

What I have found that meets what I haer as harshness is a lack of HF smoothness and an associated group delay such as occurs with diffraction and or very early reflections (< 1ms.). This is well defined objectively.

 

[LineArray]

In my personal experience, when talking about common
2 way designs e.g., even a smooth midrange having low
distortion and absence of driver and cabinet colorations
(high Q resonances) contributes to avoid harshness.

Problems in midrange seem to mislead some designs
to make use of an "overbright" adjustment, trying
fake "detail resolution".

While you can have fairly smooth response at HF using
a 30 Euro dome tweeter mounted on a canny shaped
baffle, it is far more challenging to get the midrange
"right", and you can mess it up easily even with a decent
(and expensive) driver ...

 

[bobodioulasso]

Agreed.

 

[Charles Darwin]

It's in the treble ok but it could come from the room (too reflective), the amp (clipping) or the tweeter (rising response or break up which is mostly found in metal domes IME, discontinuencies in the horn/compression driver if used).

One, two or all could be involved in any individual case of harshness.

 

[DBMandrake]

In the world of psychoacoustics there are many terms that have well defined meanings. Brightness, Roughness, etc. are all well defined, but harshness is not IMO.

Perhaps not well defined, but its something that makes you feel discomfort at anything other than a low volume level, (a bit like a scratched chalk board) triggering the "turn the volume down" reflex at much lower than normal SPL's.

What I have found that meets what I haer as harshness is a lack of HF smoothness and an associated group delay such as occurs with diffraction and or very early reflections (< 1ms.). This is well defined objectively.

I would definitely agree that harshness is a lack of high frequency smoothness, as measured narrow band with sufficient frequency resolution.

1/3rd octave measurements etc will not reveal it - a speaker can measure flat with 1/3rd octave averaging and be tonally balanced, but if the narrow band response is spiky it will still sound harsh.

Conversely, a speaker whose frequency response is not flat, but varies very smoothly and gradually (measured narrow band) can sound tonally unbalanced, yet not harsh. (Unless the treble is tilted up a lot)

Any significant narrow band peaks or dips in response at high frequencies can cause harshness, whether due to driver resonances like cone breakup, or diffraction.

What constitutes "High frequency" is hard to pin down exactly but starting somewhere from about 2-3Khz upwards. 4Khz is a particularly "sensitive" area, but I have heard harshness from fairly small resonances at 8Khz and above as well, even as high as 15Khz.

Lack of response smoothness significantly below about 2Khz doesn't seem to cause harshness in my experience.

Different songs tend to stimulate different treble resonances that a speaker might have, and as resonances are systematically eliminated by design or (sometimes) EQ "harshness" on particular songs goes away... A narrowband peak of only 1dB in the treble can be enough for noticeable harshness on particular songs that constantly hit that peak, especially if it's straddled by vibrato, yet other songs may sound perfectly fine.

If the source of the peaks and dips is driver resonances then mechanical damping and/or some EQ in the network can often fix it, if its diffraction related it can't be fixed by EQ, instead you need to deal with the diffraction.

The two can't be reliably separated on a single on-axis frequency response plot so you need to look at the impulse response and CSD, (for diffraction and resonances respectively) and/or compare the on axis and and near off-axis responses looking for the effects of diffraction, to make sure you're not trying to apply EQ to a response bump that is actually caused by diffraction.