The difference between treble response and leading-edge transient response

[markl]

This has percolating in my mind for a little while, and I'm wondering if it's possible that some of us are mixing up two separate concepts when we describe the sound we hear of any given component/headphone or even describe the sound we perceive of an individual recording.

We all know what treble is, it's the high-frequency sounds typified by cymbals, high notes on a piano, the triangle, high-pitched horns/trumpets, etc. But there's another phenomenon, and I don't know the exact technical term for it, so I'm going to call it "leading-edge transient response". By that I mean, the "sound" of the very first part of any piece of musical information, any musical event, that reaches your ears. I think I see these two concepts getting mixed up from time to time. I thought it might make for an interesting discussion...

Some components are very fast, aggressive, have a very hard "attack", a sound that leaps out at you in a very springy accelerated way. Furthermore, this leading-edge transient, in addition to it's perceived velocity, can have another characteristic, a "texture" if you will. It can be a little frayed at the edges, have some "bite" and "zing" to it, maybe be a little jagged and hard edged. OTOH, it can also be soft, smooth, and rounded, yet still be aggressive, fast and "forward". It can have a thin leading edge that builds gradually until the whole event is fully heard, or it can have a steep and exaggerated almost brick-wall-like front end that wallops the ear all at once.

The leading-edge transients occur throughout the frequency spectrum and can be caused by any musical event from a tight snap on a snare drum, to a hard blow on a trumpet, to the pluck of an acoutic guitar, even the pluck of a string on a bass guitar. All these musical events have a sort of "shape" to them, and IMO, just as with any other characteristic of sound reproduction, can be presented by any given component in any number of ways as described above.

Because these leading edge transients are so noticeable to us, I think they can get mixed up with what some people refer to as "treble response" of some components, because they can share a similar psycho-acoustic effect on the ear, we react subconsciously in a similar manner to components that are "too bright" as we do to components with an unpleasant "leading-edge transient response". We may flinch, squint our eyes, reach for the volume knob to turn it down, etc., or it may be at a level where it's just a subtle unconscious reslessness and irritation or "fatigue".

I think that with some particularly "forward" sounding pieces of gear, this added attack with its extra speed and velocity, combined with the particular characteristic of that leading edge transient (jagged, sharp, hard) can be interpreted by some people as "treble" response, as in "wow, is that headphone colored with an exaggerated treble". When in reality, they aren't reacting to its actual treble response but the nature of its "attack" and the characteristics of its texture, information that is coming at the ear from all areas of the frequency response, not just treble.

I'm not sure if I've captured exactly what I'm trying to convey, but if anyone can make anything of that and wants to elaborate or pontificate, feel free.

 

[halcyon]

The difficulty in talking about these things is that what we measure doesn't always correlate with what we hear.

As such, I think your approach of talking about the perception makes more sense.

However, being that I try to always move between the measurement and the listening domain, I will try to clarify some things that I don't think are mutually exclusive with what you are saying.

I do think that the transient response (rise time if you will) of a system will affect how we hear some sounds as coloured, muffled or powerful and stingy (or something in between).

I also think that this is not the same thing as treble detail (i.e. prominency of high frequencies in the output), leading often to perceptually to increased depth cues, positioning information and some sparkling associated with certain types of instruments (e.g. bells, triangles, cymbals).

Also, I don't think that treble detail necessariliy (or even as the key contributor) leads to listening fatigue, but uneveness in the presence region, which is in the upper mid-frequencies (c. 2.5 - 3.5 kHz).

I think transient response of the system and treble prominence of the output are different things, but lack in the capacity of either capability can perceptually "dull" the sound. Perceptually they are different, so I'm not too worried about mixing them.

However, I do believe that many people find "added details" in systems that have elevated some areas of the frequency spectrum. This can be in the treble, but mostly it happend in the middle frequencies.

Transient response capability, imho, brings forth the sheer believability of attack sounds, they can be downright scary on the best of recordings. Does this bring about added detail, more than just the feeling of dynamics? Yes, I believe it can do so, but I do think the perceptual effect is different from (linear) frequency manipulation.

 

[JaZZ]

Interesting topic. Thanks, Mark!

Theoretically -- and actually -- flat frequency response means also ideal transient response, including the «leading edge». Just like the frequency response is readable from a pulse or a square-wave response. But that's only half the story.

You never get a flat high-frequency response with headphones due to early reflections between membrane and outer ear with consequential interferences, meaning a somewhat jagged frequency response. To make it even worse, it is different with different ear shapes. Add to this that a flat frequency response doesn't even mean adequate reproduction of a natural sound field, because the pinna naturally causes considerable (FR) distortion, the more so with frontal sound events. Conclusion: you'll barely ever be able to measure the leading-edge behavior with headphones. BTW, I wouldn't pay too much attention to the mere rise time of the membrane (although it's certainly not negligeable) because of the fact that the limited bandwidth of sources (redbook CD!) and recordings (data format, tape recorders, microphones...) decisively limits its benefit.

That's the theoretical part. I think with a lot of listening experience it's possible to discern leading-edge accuracy from fake «quickness» and brightness. One of the most important physical preconditions is acoustic «dryness» (low reflectivity) within the driver, otherwise the leading edge (or «first wavefront») is smeared and loses its character. Now the acoustics with headphones make it impossible to preserve it perfectly, but I guess this wouldn't even be absolutely favorable to the sound. I mean, to have a sound source directly in front of the ears is already an unnatural situation; if it would competely lack any (reverberative) smearing effects -- which in fact are a natural function of the outer ear -- it would cause a very fatiguing sound. Not to speak of how every tiny flaw of the electronics would be magnified to the excess.

This may sound like a denial of the benefits from transient accuracy. No, it's just that it has to be relativized. It's strongly dependent of synergies, mainly between headphone and ears. But I think there are headphones which get more praises as to transient-response acuracy than others, so the relativization has to be relativized itself. To my understanding the often slightly contemned foam pads between membrane and ear contribute to signal accuracy in the sense of preserving the leading edge -- by dampening early reflections/resonances. That's why they are quite common throughout all the different brands.

 

[markl]

Quote:

I also think that this is not the same thing as treble detail (i.e. prominency of high frequencies in the output), leading often to perceptually to increased depth cues,

Halcyon, very interesting, I hadn't considered how leading-edge transient response might affect depth perception, that's definitely something else that deserves some thought... A more impuslive impetuous and forceful driver might make everything seem "closer", where a more lazy, slow and reticent driver might make evrything seem farther away... Quote:

Transient response capability, imho, brings forth the sheer believability of attack sounds, they can be downright scary on the best of recordings. Does this bring about added detail, more than just the feeling of dynamics? Yes, I believe it can do so, but I do think the perceptual effect is different from (linear) frequency manipulation.

I agree these are definitely different things, but I think in reading some people's comments here and there they are two things that are easily mixed up. I think it's an interesting question to ask, whether the subjectively "fastest", most projecting, penetrating, and impulsive sound is the most "neutral" and least "restrictive" on the signal, or whether it is possible for some equipment to be sort of "hyper-active" and over-empasize the actual dimensions and attack of the leading edge transients contained on the recording, that given even a tiny stimulus, they overshoot and over-react. My gut tells me it *is* possible for gear to exaggerate this effect, but I'd like to hear from others on this..

JaZZ,
I wish I understood more of your post so I could comment intelligently...

Thanks for your thoughts, hopefully some others will chime in...

 

[JaZZ]

Quote:

Originally Posted by markl JaZZ, I wish I understood more of your post so I could comment intelligently...

Oops... sometimes I have the suspicion that my wording isn't quite from this world...

What I tried to say: An infinite bandwidth is synonym to ideal transient response and infinitely short rise time; a treble roll-off means rounded transients and increased rise time; a nonlinear frequency response means a screwed up transient response.

A headphone by nature (because of reflections/resonances/interferences) doesn't allow a flat frequency response, so the transient response is automatically in some ways crippled. Moreover the reverberative behavior of the sound waves within the headphone/ear system causes a smearing of transients. Both phenomena heavily affect the accuracy of the leading edge a priori.

What I also wanted to point out is that under normal, non-headphone-listening circumstances the outer ear by nature has a transient-smearing function. So with headphones, where the outer ear is at least partly bypassed, a noncompromising avoiding of transient smearing could even lead to a fatiguing, dry sound which would magnify electronics flaws.

Nevertheless, the driver's quality in terms of transient (and leading-edge) response seems to have some importance anyway. I agree with halcyon in his speculation about the meaning of transient response (and here especially leading edges) for the perception of spatial depth: the precise reconstruction of space-relevant information within the signal facilitates the differentiation between direct sound and reflected, diffuse sound as well as to decode runtime informations for left-right localization.

In my perception headphones with more upfront characteristic and closer soundstage are such with a more present treble range -- and vice versa. The perceived distance isn't directly related to transient response. There are even cases where bad transient response and consequential bad resolution kill the spatial information in the recording and thus lead to a flat, very close soundstage. Another case are supraaural headphones where the total bypassing of the outer ear leads to a more direct sound with lower diffuse sound content.

 

[PedigreeE]

Sounds like a CD3K defense, hehe, I love mine don't get me wrong. Very interesting thoughts though, my head hurts thinking about the shape of sound, while I’m at it I might as well ponder the space time continuum.

 

[Dusty Chalk]

What JaZZ is saying isn't entirely correct. First of all, Mark, you're absolutely right that they are two different things. But what JaZZ is getting wrong is that not everything can be mapped to frequency response. The thing you're talking about with transients is more important in the phase response. Now, what I think JaZZ is talking about is that when one does a proper FFT analysis of a signal, technically one should be incorporating phase into the calculations. But measuring something using FFT analysis and describing its characteristics in the frequency domain are not the same thing. There's also the time domain, and that's the differentiation I think you're trying to make.

Now, that said, there is a relationship between the two, as has already been aluded to. If you fail in one, you're most likely going to fail in the other. But succeeding in one (especially in, for example, speaker design, or, more specifically, crossover design) does not guarantee succeeding in the other.

The problem is that, when one measures something's frequency response, then one usually measures over a course of a period of time. That's why waterfall plots are so useful (but, like any other plot, never covers all of the information), for it shows both.

Anyway, don't mean to trash JaZZ' statements, there is some good stuff in there, but I just didn't want there to be any confusion between a good or adequate frequency response -- especially in the treble area -- and a good transient response. Not the same thing.

 

[JaZZ]

Quote:

Originally Posted by Dusty Chalk What JaZZ is saying isn't entirely correct. ... I just didn't want there to be any confusion between a good or adequate frequency response -- especially in the treble area -- and a good transient response. Not the same thing.

Yes: Flat frequency response means perfect transient and perfect phase response -- at least in the case of fullrange sound transducers. Phase distortion in the form of group delay with at the same time flat overall frequency response can only appear in multiway speaker systems. In turn you never get perfect transient response with a distorted frequency response. At the same time the latter also means phase distortion. Of course perfect frequency and transient response is only a theoretical assumption for illustration purposes; in reality you always have to deal with limited bandwidth and all kinds of distortion, but they always go hand in hand in the amplitude and time domain.

To make things clear: Flat frequency response in the example means infinite bandwidth! Only this can also guarantee perfect transient response. So a flat frequency response e.g. just within the audio band does by no means guarantee perfect transient response.

 

[Catharsis]

Rise from the dead thread!

This is ultra-interesting. I think I understand what Jazz is getting at. Perhaps the supra-aural design also explains (in part) as to why Grado headphones are so snappy!

 

[Catharsis]

And not to overlook the numerous interpretations of speed by so many helpful head-fi members, professional definitions of transient response (from sound engineers), have also suggested that a flat frequency response IS the end result of numerous factors, one of which is good transient response.

When people suggest that driver acceleration is equivalent to transient response, I can't help but think of a driver responding to a high freqeuncy input:

the driver must accelerate from 0 to a a wave-peak....stop.....travel the entire distance of the sound amplitude to reach the wave-trough...stop.....and repeat in an oscillatory fashion 20,000 times in a single second. That's from 0 to full peak to stop and reverse in 0.00006 seconds. That's some serious stop and start action and it all takes place 20,000 times in a single second at a volume that is properly relative to the other frequencies. That's f#$%ing fast.

Echoing jazz, a flat frequency response both low extension and high extension would NEED to be "fast" in order to obey the laws of physics.

 

[b0dhi]

Quote:

Originally Posted by JaZZ /img/forum/go_quote.gif Yes: Flat frequency response means perfect transient and perfect phase response -- at least in the case of fullrange sound transducers.

As has just been pointed out 5 years ago, flat frequency response does not equate to phase linearity. Though impulse response can be used to determine frequency response, it doesn't go the other way. One captures time and frequency domain information, the other captures only frequency domain information.

 

[JaZZ]

Quote:

Originally Posted by b0dhi /img/forum/go_quote.gif As has just been pointed out 5 years ago, flat frequency response does not equate to phase linearity. Though impulse response can be used to determine frequency response, it doesn't go the other way. One captures time and frequency domain information, the other captures only frequency domain information.

A flat frequency response can't have phase distortions – think about it!
.

 

[penguin121]

Quote:

Originally Posted by JaZZ /img/forum/go_quote.gif A flat frequency response can't have phase distortions – think about it! .

When you transform time data into spectral data, the frequency response function you get will be complex (have amplitude and phase). However, frequency response is typically plotted in dB, which only gives you the amplitude of the frequency response, basically throwing out half of the information from the original time domain signal. In order to recreate transients properly, the proper phase relationship is required. Since the dB frequency response doesn't tell you anything about the relative phase of original time signal, it is not sufficient information to describe transient performance. Obviously bad frequency response will cause problems with transient response. However, good frequency response by itself will not guarantee good transient response.

 

[JaZZ]

Quote:

Originally Posted by penguin121 /img/forum/go_quote.gif When you transform time data into spectral data, the frequency response function you get will be complex (have amplitude and phase). However, frequency response is typically plotted in dB, which only gives you the amplitude of the frequency response, basically throwing out half of the information from the original time domain signal. In order to recreate transients properly, the proper phase relationship is required. Since the dB frequency response doesn't tell you anything about the relative phase of original time signal, it is not sufficient information to describe transient performance. Obviously bad frequency response will cause problems with transient response. However, good frequency response by itself will not guarantee good transient response.

I agree. But «good» frequency response is not perfect frequency response. I was talking of the latter – from a merely theoretical perspective.

However, a fairly flat frequency response will guarantee a fairly good phase and transient response. (Still with respect to full-range transducers.)
.

 

[b0dhi]

Quote:

Originally Posted by JaZZ /img/forum/go_quote.gif I agree. But «good» frequency response is not perfect frequency response. I was talking of the latter – from a merely theoretical perspective. However, a fairly flat frequency response will guarantee a fairly good phase and transient response. (Still with respect to full-range transducers.) .

I'm curious as to why you believe this?