'Flat' is not correct for a stereo system ?

[gainphile]

SL recently wrote:

Experiences with equalizing an initially flat on-axis response to obtain the realistic sounding ORION-3.2 have convinced me that 'flat' is not correct for a stereo system. Here we must create believable phantom sources from two real sources at +/-30 degree angle to the listener, at the 'sweet spot', in the room. Nor is the BBC dip the answer. Instead, the response must be flat above 100 Hz and then gradually drop to a lower level at high frequencies. There are general physical and psychoacoustic requirements for such a response, which I will explain at a later time. A specific on-axis frequency response curve is usually the secret behind a good sounding loudspeaker. The marketing department likes to label this curve 'flat'.

Linkwitz-Links

I'm about few thousands kilometers from my speakers so I have yet to build and see/hear this curve for myself. But is it reaaally ??

[Saurav]

Another discussion on this subject:

Psychoacoustic considerations for my OB project

[wintermute]

Interesting!! This tends to fit with my preferrences for a slightly subdued treble response! I find that when I adjust for flat treble response I don't like the sound, when it is slightly more subdued than 'flat' I find the speakers much more enjoyable

Of course since my microphone isn't calibrated (WM-60AY so pretty flat usually) it is hard to say whether what I've measured is a true representation....

Tony.

[cuibono]

I like to roll-off my treble by a dB or so too. In the thread above, JohnK says he does something similar. I think others do too, and I strongly suspect it has to do with the room one is listening in, and personal preferences.

One should note that there is a backstory to SL's observations, that is discussed in the above thread.

[Cal Weldon]

Perhaps there was a wee too much time on the tools before I leanred the value of ear protection so while I agree to a point on the downward FR slope, I actually need a boost above 10K.

I no longer hear above 14k so compression drivers are just fine thanks.

[speakerdoctor]

This one's over 300 posts already. Good stuff....

Flat Response - Techtalk at Parts-Express.com

[Saurav]

By "roll off the treble", what are you doing exactly? Setting the tweeter level lower? Using a shelving filter like SL? Something else?

[john k...]

Originally Posted by gainphile

SL recently wrote:

Experiences with equalizing an initially flat on-axis response to obtain the realistic sounding ORION-3.2 have convinced me that 'flat' is not correct for a stereo system. Here we must create believable phantom sources from two real sources at +/-30 degree angle to the listener, at the 'sweet spot', in the room. Nor is the BBC dip the answer. Instead, the response must be flat above 100 Hz and then gradually drop to a lower level at high frequencies. There are general physical and psychoacoustic requirements for such a response, which I will explain at a later time. A specific on-axis frequency response curve is usually the secret behind a good sounding loudspeaker. The marketing department likes to label this curve 'flat'.

Linkwitz-Links

I'm about few thousands kilometers from my speakers so I have yet to build and see/hear this curve for myself. But is it reaaally ??

I think at this point you have to take some of what SL says with a grain of salt. This argument about flat on axis response has a lot more to do with what is happening off axis and total power response as well as room acoustics. SL's comments are made with respect to the current "tuning" of the Orion. I believe a lot of this has to do with the mismatch of the polar response above the crossover from dipole mid to dome tweeter. I see the same thing in the NaO II. In a live environment it can sound a little bright which is why the NaO II has always had a tweeter level control.

I recently did an experiment listening to the NaO II side by side with the NaO Note. As you may be aware, the NaO Note maintains dipole response to a much higher frequency than either the NaO II or the Orion, and above 4k hz the polar pattern begins to narrow. You can view the polar response of the NaO II here and Note here. When I initially listened to the Note beside the NaO II the note sounded different, more so than I expected. The NaO II has a highly detailed and somewhat etched sound to it, much like an ELS. Not surprising as I used my Martin Logan Monoliths as a reference. (One of the goals of the NaO II was to sound similar to the Monolith but without the large panel beaming.) Thankfully, the Note sounded more natural, still highly detailed but not so ELS like. Still I wanted to understand the source of the difference. So what I did was to digitally equalize both main panels to have a 100 Hz LR4 high pass response which was ruler flat on axis and also matched the on axial levels. Thus all that remained was the effects of polar response. The difference between the NaO II and the Note remained. But what was interesting was that I could make the Note sound very much like the NaO II by boosting the response above 3k Hz. Similarly I could make the NaO II sound similar the Note by shelving down the response above 3k Hz. What this tells me is that at higher frequency it seems to be more about power response than on axis response. Thus messing with the on axis response at higher frequency seems more to be compensating for the behavior of the polar response. It is not so much that the on axis response should droop at higher frequency. It is that the power response should. This can be achieved by keeping the on axis response flat while narrowing the polar response. In the case of the Orion, my opinion is that SL is actually compensating for the broadening of the polar response above the crossover by shelving down the tweeter level.

I don't think this should come as any surprise. After all, take a speaker that sounds good in a highly damped environment and place it in a highly reflective environment and what do you do? Turn down the treble.

SL's comment, "A specific on-axis frequency response curve is usually the secret behind a good sounding loudspeaker" has more to do with what the polar response of the speaker is than what the on axis response is, IMO. Otherwise all we would need to do is measure the on axis response of that mythical good sounding speaker and all speakers with the same on axis response would sound the same. Obviously, if the on axis response must be tailored based on the speakers polar response to yield the correct power response then the inverse should also be true; for a given power response we should be able to tailor the polar response so that the on axis response is flat. This is what I have attempted to do with the NaO Note. As you may be aware I have been adamant about trying to correct the discontinuity in the polar response of speakers like the NaO II and Orion for some time. I believed the solution was to start with a clean sheet of paper which is what the NaO Note is all about.

[lbstyling]

But where do you fit in people who like the trebble a little hot (like me)?? I usualy find a flat FR a bit lifeless.

ive had my ears tested and was told my hearing was exceptionally linear and good to 19khz. (Im 25)

[Drew Eckhardt]

Originally Posted by gainphile

SL recently wrote:

Experiences with equalizing an initially flat on-axis response to obtain the realistic sounding ORION-3.2 have convinced me that 'flat' is not correct for a stereo system.

Correlation does not imply causality. I think it's more likely that we're looking for a reduction in the total direct and reflected energy towards high frequencies which can also come from flat-on axis response with monotonically increasing directivity.

IIRC Toole and Olive's research suggests that our brains form their impression of timbre as a combination of the direct sound and what we identify as reflections.

That explains empirical observations of speakers with flat on-axis response and broadening polar response towards high frequencies being perceived as bright which can be tamed by non-flat response.

Since the Orion front hemisphere polar response broadens from 800Hz - 4KHz, it should need some on-axis attenuation.

As a counter-example, Pluto works great with flat on-axis response until the last octave and monotonically increasing directivity.