Do we REALLY want a flat response? (1 Viewer)

[Bill Catherall]

I was browsing through my June edition of the Evaluation Engineering magazine (I get it free at work) when I stumbled on an article about Link Removed. The article displayed Link Removed and talked about how our ear sensitivity isn't flat and is dependent on both volume and frequency. The way to interpret this chart is basically a 100 Hz signal will need to be played at 65 db in order for it to sound as loud as a 5 kHz signal that is being played at 50 db. But as you increase the volume the perceived loudness flatens out across the lower frequencies and our hearing becomes more sensitive.
But even at movie reference levels the mid to high frequencies seem as much as 10-20 db louder than the low end. So while our speakers may measure a perfectly flat response using instrumentation, our ears don't hear a flat response. Then when listening to music, which we typically do at much lower levels (because it's hard to carry on a conversation with loud music), our hearing sensitivity drops drastically for the lower frequencies.
Am I misunderstanding this? Shouldn't speakers be designed with this in mind? It seems we should really want the speaker response curves to match those found in the chart above and not be flat.

 

[Vince Maskeeper]

Bill,

I spoke about this concept a little bit when the issue of the "house curve" first popped up on these forums. The concept of a Fletcher-Munson curve to comensate at lower volumes is one issue- however the idea of completely compensating natural human hearing is another...

First, let me say that the info you have cited is indeed correct- that the human hearing is not flat- and it's degree of "not flat" does vary slightly with different volume levels. We have better hearing response to midrange audio than we do to high and low freq. So your immediately thought would be that we need to boost these freq in order for it to sound proper to us.

However- you are omitting an important factors. While you are correct that our hearing isn't flat- neither is the material we're listening to. If you were to examine on an RTA a direct feed from a pleasing piece of music- you would find the time honored "smiley face". The material itself is already scooped in the midrange and peaked at the outer low and high freq.

Man naturally creates material that is pleasing to our ear- so by it's very nature is is not flat. The natural freq response of nearly anything you listen to on a daily basis has already be compensated for your hearing in their creation...

Remember that the materials you're listening to were prepared by a human being. Therefore, the curve of human hearing is already built into the material. Certainly there can be some argument in terms of how close your version of "human hearing" is to the engineer who prepared the material, but I think that is beyond your question.

The point is that we do not prepare a flat system in order to produce flat material. Instead- we prepare a flat system in order to create a flat starting point-- to emulate the environment the person who prepared the material experienced. Because the preparer also did not have "flat" hearing- the entire nature of the finished recording will not be flat- rather specifically tailored to sound good to a human being's response curve (depending of course that it is played back on a system with the same starting "playing field" as it was prepared, in other words being flat).


So the bottom line is by preparing a system to be "flat"- you do not get "flat" output in an absolute sense (this would not be at all appealing to our ears). Instead you get a "flat" system basis in a relative term from which to then reproduce the finished material.

-V

 

[Roger Kint]

I see you've stumbled across the Equal Loudness Curve chart. It's true our ear sensitivity isn't flat across intensity and frequency spectrums and that's why some components have a 'Loudness' button to compensate for lower volumes.

If speakers were designed to compensate for the ELC, then it would be off at higher volumes.

Add to the fact of ELCs the inevitable loss of hearing at older ages and also the fact that many people hear differently(one may be ear more sensitive than the other), many or all speaker companies would be better off pursuing a flat freq response.

 

[Michael Lomker]

Not really. Live instruments do not adjust their levels with human hearing in mind. Our equipment is designed to reproduce reality, not to improve upon it.

The other factor is that such adjustments could be EQ'ed in by a sound engineer when it is recorded. If the speaker essentially made EQ adjustments to the sound then you'd never know what the music was intended to sound like.

 

[John F. Palacio]

"Am I misunderstanding this? Shouldn't speakers be designed with this in mind? It seems we should really want the speaker response curves to match those found in the chart above and not be flat."

I think you are. Whether speakers should be flat or not is a topic of hot debate. But we do not really want the speakers to compensate for the Fletcher-Munson effect at a particular loudness level as this will make them "wrong" at any other SPL.

In the same vein it should not try "inverse compensation" for the other non-linearities of the human ear.

This was done eons ago when man started to make music. The tonal balance of different musical instruments as well as the combination of several instruments was made to please the ear.

The recording-reproducing chain should strive to archive and then present it to the listener.

Whatever we hear in listening to a live band, for instance, should be captured and time-shifted.

IMHO

 

[Bill Catherall]

Oh yeah. I guess if the speaker did its own EQ then it would color the sound differently and tweek the "raw sound" so it sounds differently than if it were live.
Sorry about the brain fart. Thanks for the adjustment guys.

 

[Holadem]

Remember that the materials you're listening to were prepared by a human being. Therefore, the curve of human hearing is already built into the material.

This nails it perfectly.
--
Holadem

 

[Martin G]

Not really. Live instruments do not adjust their levels with human hearing in mind. Our equipment is designed to reproduce reality, not to improve upon it.

Actually live instruments adjust their levels all the time. I have been playing horn for about 14 years now and we are always taught that you play lower notes a little louder then higher ones. This is true for large music groups as well. You always try to follow the "balance pyramid" where the lows are louder and the highs are the softest. Otherwise things sound unbalanced even if they are all playing at the same decibel level.

 

[Dan Hine]

Martin is right. While the instrument itself may not be made to adjust its levels, the musician is taught to do just that. It's all about playing in ensemble. At least it has been in my 8 years of playing. Well, actually 9 but the first year doesn't count. I had a shiny metal thing that made noise...what do you mean I don't get to play loud?
Dan Hine

 

[Mark Fitzsimmons]

I had a shiny metal thing that made noise

Triangle?

 

[Dan Hine]

Trombone...but you get points for the correct first letter.


Dan

 

[David Dranove]

There is another important ingredient -- the room. Some speaker manufacturers account for room acoustics when they design their products. I own a pair of Avalon Arcus speakers. Avalon designed the port on this speaker to generate a modest amount of bass energy down to 20 hz or so. In doing so, Avalon recognized that many (most?) listening rooms have room modes in the 20-40 hz range. When there are room modes, a little bass energy can go a long way. In fact, I get nearly flat bass response in my room down to 30 hz, with audible bass down to 25 hz, despite a bass output curve from the speaker that drops off below 45 hz.

If my speakers were "flat" down to 20 hz, then there would be a noticeable spike at 25-40 hz, a spike that no amount of bass traps could eliminate.

 

[Mark Seaton]

We are trying to present the ear with the same sound that originally occurred. We don't want to match the equal loudness curve, but rather recognize the change in the curve as level changes. The idea is that if two people hear one instrument playing at the same time, and you then reproduce the sound exactly both will agree that it is the same as the original instrument. They may have different oppinions as to the character of the sound, and their preferences, but the likeness to the original is what we are striving for.
Regards,

 

[Chris Tsutsui]

"Our equipment is designed to reproduce reality, not to improve upon it."

I think that the point of music is to produce sound exactly how it was intended. To do this there has to be a way to measure and match the 2 sounds. Frequency response is a tool used to record those sounds on paper.

Some people like to listen to speakers that sound euphonic to their own ear. Then there would need to be adjustments to the sound curve in order for the original recreated piece to sound good.

I'm not sure how good measuring tools are at comparing sound but I'd imagine they will improve over time. Perhaps a good AB comparison would be listening to a single vocalist sing in a sound room, and then switching to 2 speakers to the left and right of them which produces the same song right where the real vocalist head is.

 

[Wayne A. Pflughaupt]

Remember that the materials you're listening to were prepared by a human being. Therefore, the curve of human hearing is already built into the material.

This is a nice theory, but anyone who has more than a half-dozen CDs or DVDs knows that the frequency balance of program content is all over the map. Am I the only person here who has some CDs and DVDs that are bass-heavy, others with a bad case of “where’s the bass?” syndrome, and still others that sound overly bright?

Unfortunately, a studio control room bears little similarity to the home environment. Not to mention, much (if not most) studio mixing is done with the speakers very close to the engineer. All a recording engineer can do is make his product sound as good as possible, but he ultimately does not know if you will be playing his CD in a car, a small bedroom, large living room or a huge disco club.

Regards,
Wayne A. Pflughaupt

 

[jeff lam]

This is one and probably the only issue I dissagree with Vince on. I tend to agree more with Wayne in this issue. As most of us don't listen to music and movies in a recording studio like the recording engineers do, our rooms are very different. We also may listen at levels very different from the recording engineer. What I have found to work well in my room and my prefered listening level is, use some of the music Wayne suggested to create your house curve(walking bass lines, bass slides, sweeps, etc.) and work on your eq till the tones become fluent and flow together without loud peeks or valleys. You must however set your prefered listening level first as the bass response changes as volume is increased. For instance, a low listening level of 60dB will require a much steeper house curve than a listening level up around 90-100dB.
I have found that I like this sound much better than a instrument measured flat response.
A ruler flat measured response sound just that... FLAT.

 

[Vince Maskeeper]

I feel as though I may have been misunderstood on this issue, given the above statement, so I'm hoping to clarify:

1) I really have nothing against the concept of a "house curve". I think it's a dandy idea. I also have nothing against EQing a system to create even response- in fact it's been a big thing I've suggested since I first popped up here a few years back. Coming from pro sound environment- EQing is a daily part of what I do, and have always carried a lot of those ideas into the hobbiest HT realm.

2) I am not suggesting that because a recording engineer prepared a masters, everything will always sound great. What I was suggesting was that some of the basic characters of human hearing will be built into the master as a result, so we don't need to completely re-tune the system to cut/boost what humans hear/don't hear.

An example being- if, let's say, all humans had a spike in hearing perception at 800hz. A flat tone, because of our perception would sound like a giant honk at 800hz. If this was true, chances are that not many instruments would contain excessive amounts of 800hz, because we would not find it pleasing. In turn, not many recordings would contain 800hz, because the engineer who prepared them would find it as displeasing as everyone else.

You can see this type of thing in most recordings. Most roll off heavily around 14k, because we don't find this excessive high end pleasing. There is no real need to dial all the 14k+ freq out of your system, because an average recording would already be naturally absent of much of that material.

I'm not really sure what these responses are actually disagreeing with-- so I must have been unclear in my initial post. All I said above was that most recordings, because they're produced by humans, will have a basic compensation for human hearing built in. I did not and would not suggest that you aren't free to season your own system to your liking or that you will like all the material an engineer will produce- however I simply responded to the original query about the need for retooling speakers to compensate for the human hearing curve.

-V

 

[Kevin C Brown]

I am always intrigued by questions about "flat frequency response" and whether that should be the goal or not.

I believe, that at the very least, you should try to *start* with a flat frequency response, and then *compare* changes made to that "benchmark". At least then, you have a constant way to compare *different* changes.

And then as well, somehow you have to take the room into consideration. Room induced peaks and valleys, "live" vs "dead", etc.

 

[Shawn Solar]

When I got my BFD I tried to eq it flat using the correction values for the spl metre. I found I prefered the bass when I did NOT use the correction values and adjusted for a flat response. I was under the belief the spl metre was designed to mimic the human ear. I know an spl metre is not the weapon of choice and is not as acurate as a RTA. It sounded a lot smoother and blended with my main speakers a lot better. It didn't sound as if the bass were missing like when I used correction values.
I also thought Dolby digital compensates the sub bass and raises the lfe level 10db(hence the 115db reference level) I am not sure if that is just headroom or it is mixed 10db higher too.
Measuring is important but if it doesn't sound right when its all done than it is useless. I always found that you measure to find a benchmark then adjust to taste. Almost like baking

 

[Saurav]

I believe, that at the very least, you should try to *start* with a flat frequency response

Except when that throws other parameters out of whack. For instance, you could add an equalizer into the signal chain and get the flattest frequency response you want, but this would probably negatively impact the system's phase response (though modern EQs are pretty good at leaving phase unharmed).

Frequency response isn't the only parameter to measure a system's accuracy, and IMO it's not the most important either. Sometimes people seem to get caught up with the quest for flat frequency response, and lose sight of the fact that it might come at the expense of something else that's equally important. This is true while designing a product, all the way to buying that product and setting up a system.