Good Treble = Detailed. No, really. - Page 3

Please though lets try to keep this discussion a matter of science rather than of opinions and how we "feel" about it.

Still, I'm fairly certain we can define in terms of the sound signal reproduction what we mean when we say "detailed". Once such a definition is reached, we can analyze it logically.

For example "siblance" as defined as strong harmonic overtones creating a piercing sound is a rather well understood concept mathematically (nonlinear signal transformation produces harmonic overtones at edges of the devices operational frequency band).

Certainly headphones are going to be more complex physically than just a frequency response graph, but still I don't think there's any escape from the fact that poor treble equates by definition to poor representation of micro-details in the sound. Regardless of how one accomplishes "good treble", good treble = detail nonetheless. Again don't mistake this for "emphasized treble", which is not what this thread is about.

For example my HD650s from my amp does not have particularly good treble, simple as that. If somehow a better amp gave better treble, even if it was recessed, that would still result in more detail in a subtle way. Treble = detail, even if that treble is recessed. I've verified this experimentally as well. My SRH940 has excellent treble, and still sounds twice as detailed as my HD650 even with the treble equalized down to comparable levels.

The argument then could be said to be: You cannot have a detailed sound if you remove treble, and therefore when it comes to detailed sound, it's largely the "good" parts of the treble that matter.

The problem with applying a low pass would be it would have to be a really good true low pass filter... that is, to completely eliminate treble, for a fair test. Because a headphone with good quality treble will still sound detailed with reduced treble volume, just like a headphone with muddy treble still sounds muddy when you emphasize the treble.

The reason I say this is with my EQ, if I completely slide down the treble region, I can still hear it clear as a crystal because it's still there, just less of it. But I don't have the greatest equalizer on my sound card, so you may find something better.

As to how to define good quality treble, I'd say a fast accurate transient response would fit the description. In any case I don't see why people are getting so upset regarding the definition of "good treble".

Also, do people realize that transient response values of headphones are going to be on the scale of ~25 microseconds (~40 Khz, has to be at least twice human hearing frequency by Nyquist-Shannon theorem) to be worth anything meaningful within human hearing. If you try to apply this in a meaningful sense to bass (say, 50 hz, or 10-20ms period), you'll notice that this is on the order of thousands of times slower than the transient response. Simply put, transient response doesn't much at all for bass. Steady state error could matter if it's really really bad.

> Really, if your theory is correct ac500 (and I think I understand what you mean now), there wouldn't need to be a treble signal. The headphone would just have to be able to respond to treble. As in, it would have to be able to reproduce a clean high frequency wave, if one is present in the signal. Doesn't have to be present for the headphone to accelerate as needed. Hence why we'd need to mess with the headphone driver, not the signal.

Yes you are starting to understand what I'm saying, but I may not be explaining properly the frequency domain. Mathematically and physically it is impossible for a headphone to respond to treble without it having... well, treble. Does that make sense?

In other words, the headphone cannot magically recognize "Hey this fast changing signal here is a piercing treble tone, I'm gonna suppress it! Ohh now this fast changing signal here is rather actually a part of a detailed mid-bass signal, I better not remove the fast signal changes this time!" I can't make the distinction, so fast change in signal = fast change in signal, simple as that. And you can mathematically prove this, formally, that you do in fact need to be able to reproduce a clean treble component in order to accelerate fast as needed for other signals.

Also messing with the signal vs. messing with the driver doesn't really matter except messing with the signal changes it in a more predictable way, while the driver is a vastly more complex transformation on the original signal since it's a nonideal real-world object pushing air molecules around. The same principles of signal processing still apply in general though.

Anyway, like I said I could write up a formal mathematical proof regarding treble and transient response if you'd like, it's not hard. If anyone who sees what I'm getting at regarding frequency domain, treble, and transient response sees any errors in my thinking here, please correct me.