There must be all sorts of diffractions occurring in these horns, much like the megaphone effect on sound of well, megaphones.A way to minimize edge diffractions and also first reflections is to make a large synergy/unity horn. My friend Legis has a huge pair of those, assisted by tapped bass horns and various supertweeter/backside tweeters. The sound is like putting earphones on, even they are in a 3x5m room with concrete walls!
Legis' Horny Tales
Spheres IMO are best for reducing diffraction as much as possible. Diffraction, either within a cone or horn, and/or caused by an enclosure itself with multiple secondary sound sources (diffracted) points, changes the tonality. Think playing a flute in free air vs. playing through a horn, it will sound different. Sing through a horn (the horn aligns with your mouth) vs. free air, will change how you sound. Also, Olsen's research on the sphere showing it had the least effect on FR as an enclosure. The FR variations of enclosure types of course create a different tonality based on the type of instrument along the FR spectrum and where along the spectrum the FR is boosted or depressed from "flat".
Having said that, a change in tonality might not sound bad to a given listener based on preference and/or the natural FR of his/her ears (those Fletcher-Munsons). As you age and your high freq. hearing lessens, a boost in the upper freq. might be preferable.
Those fletcher-munsons curves are often misunderstood!
They are : Equal-Loudness-Contours (ELC)
Natural FR is HRTF , depending on angle in all XYZ directions!
In all those years there was no one who told the people where the deep 8kHz dip in the ELC is coming from!
Still waiting for a mastermind to explain on that ....
They are : Equal-Loudness-Contours (ELC)
Natural FR is HRTF , depending on angle in all XYZ directions!
In all those years there was no one who told the people where the deep 8kHz dip in the ELC is coming from!
Still waiting for a mastermind to explain on that ....
An article on HRTF...
https://hearingreview.com/hearing-p...nal-spatial-hearing-across-hearing-aid-styles
The key point is that human hearing is most sensitive in middle frequencies, and at lower amplitudes you hear these easier than an equal amplitude of bass and to a lesser extent treble. That's been well established although there are different curve by different researchers.
https://en.wikipedia.org/wiki/Equal-loudness_contour
One question I often ask is the effect on how much one's ear flap sticks out on perceived FR? If you cup your ears, "loudness" and tonality changes as increased gathering of sound waves by your hands is not uniform for the frequency spectrum. But one's ears may be very flat on the head while others stick out a lot, and in between. But many other factors in hearing also:
https://courses.physics.illinois.ed...Notes/P406POM_Lecture_Notes/P406POM_Lect5.pdf
Last edited:
The 8k dip has been a bit of an unexplained anomaly, but there are some theories stating it has to do with the circumference of the ear and how it couples to the ear canal. IMO this is why some people hear an imbalance with otherwise flat measuring full sized headphones, but find IEMs to sound more natural. It also depends on how far the drivers are located from the ears on over the ear HPs, which is why the ear cushions alone can impart a big change in perceived FR with this type of HP.
Wow! The one and only crosses my way ... can I catch you for a moment in time and space? :
At any angle (afaik) the HRTF is never flat but has certain ups and downs in FR for building a signature on it's own!
From birth on , you cannot escape this signature and your braineurons making the best out of it to gain information about the environment .
So far so good but the natural conclusion is for me - even if your hyperbest speaker has the flattest FR , is it still coloured by the natural HRTF?
So when you apply for instance a broadband correlated white noise signal to both speakers (mono) and the FR is measurable flat , you may think it sounds flat (your analyzer shows you it's flat) but thuth is , it is NEVER flat , because of HRTF and ELC and probably more brain processing due to complex input signal envelope!
Things getting even more complex if you look at the input signal using an oszilloscope and some piece of music! The closer the notes are things get worse to analyse for the EBS (Ear-Brain-System) ....
Conclusion is - a flat line on the screen is nice but only a gimmick!
An even funnier observation is : Start a wave editor program like Audacity or Goldwave that shows you the actual waveform (this is the actual content that your ears are getting) of the music that is running ... please notice IT IS NEVER FLAT !!! (not even in the shortest moment)
The flat line thing (FLT) is more a sales hype to make money go wandering around
bye!
At any angle (afaik) the HRTF is never flat but has certain ups and downs in FR for building a signature on it's own!
From birth on , you cannot escape this signature and your braineurons making the best out of it to gain information about the environment .
So far so good but the natural conclusion is for me - even if your hyperbest speaker has the flattest FR , is it still coloured by the natural HRTF?
So when you apply for instance a broadband correlated white noise signal to both speakers (mono) and the FR is measurable flat , you may think it sounds flat (your analyzer shows you it's flat) but thuth is , it is NEVER flat , because of HRTF and ELC and probably more brain processing due to complex input signal envelope!
Things getting even more complex if you look at the input signal using an oszilloscope and some piece of music! The closer the notes are things get worse to analyse for the EBS (Ear-Brain-System) ....
Conclusion is - a flat line on the screen is nice but only a gimmick!
An even funnier observation is : Start a wave editor program like Audacity or Goldwave that shows you the actual waveform (this is the actual content that your ears are getting) of the music that is running ... please notice IT IS NEVER FLAT !!! (not even in the shortest moment)
The flat line thing (FLT) is more a sales hype to make money go wandering around
bye!
Flat sounding speakers aren't flat at all. Yes that's thanks to our ear and how it couples to the air pressure changes based on our own unique neurological hearing response and anatomy/shape of our ears. I can hear flat response with the exception of.a 2dB peak error at 2 to 2.5k. That error is part of what we call the Fletcher Munson curve, which is UNIQUE for EVERY SINGLE PERSON. Some people can't hear lower than 20 Hz with their ears (not their body). Some people can't hear past 10k when their young and others hear past 17k being in their 50s. I'm over 50 and can hear to 16k down to 15 Hz. Fact is no one can perceive acoustically (not electrically) flat FR and only from about 300 to 8k. The Fletcher Munson dip is where people will disagree. Some of the FM curves show the 8k anomaly blip. Its not always factored into alot of the calibration curves for PA speaker arrays and that is a big problem. I always account for it in all of my calibration work, plus I have my own signature FM curve that I've derived with testing on a broad spread of people.
The idea of "flat" speakers is sound (pun intended.)
What we want is for the sound field on playback to be the same as that at the recording mics. This is simply a "flat" transfer function.
Ear response has nothing to do with the issue. We hear the original sound field with the same ears that we hear the playback. This takes the ears out of the picture.
There are indeed sound ways (there I go again) to accomodate high frequency perception in small rooms depending on speaker Q, but those are secondary issues that only have subtle effects from "on axis flat". There is also the LF perception issues of low LF reverb in small rooms, another secondary effect that is correctable with a subtle slope.
What we want is for the sound field on playback to be the same as that at the recording mics. This is simply a "flat" transfer function.
Ear response has nothing to do with the issue. We hear the original sound field with the same ears that we hear the playback. This takes the ears out of the picture.
There are indeed sound ways (there I go again) to accomodate high frequency perception in small rooms depending on speaker Q, but those are secondary issues that only have subtle effects from "on axis flat". There is also the LF perception issues of low LF reverb in small rooms, another secondary effect that is correctable with a subtle slope.
I do agree with you that speaker(s) as a reference source should "measure" flat in order not to impose FR errors in the accurate reproduction of a recorded source done through equally "flat" calibrated mic(s). The perception of flat FR only holds true if both recording and playback are done at the same SPL.
At varying SPLs of playback, the "flat" sound we hear isn't flat at all (as you know). This is the error I talk about, along with the 8k anomaly. A flat recording at a given reference level will obviously only sound flat played back at the exact level it was recorded at, mainly due to the FM response curve of our hearing.
As soon as you change the reference level of playback, you need to EQ the playback system to compensate for the non-linear hearing curve, otherwise it won't sound flat.
Another major contributor to how flat something sounds (and how fatiquing it is to listen to) lies in the addition of non linear distortion spectrum components contributed by the playback speaker system itself. In this case, a PA calibrated to sound "flat" won't do so in reality because the distortion added into the playback sound will give the ear false cues in subjective linearity.and loudness. This is aside from the delta reference playback level induced FR error and the complex way it mixes with the addition of distortion components. Often times, the reason we find a speaker to sound forward isnt becuase it has a peaking midrange FR, but rather a distortion profile with a higher concentration of THD in that frequency range. You can mask some of this by means of EQ, but it won't completely fix the perceived error in balance by the ear.
What measures flat in an anechoic chamber does not measure that flat in a real room at the listening spot.
Subjective preference for steady state room curves according to Toole and Olive:
Still depending on the directivity of the speakers used and the room itself though. Though I do believe it is wise to set
and use a specific SPL listening level (on average) when looking for these kind of preferences.
Subjective preference for steady state room curves according to Toole and Olive:
Still depending on the directivity of the speakers used and the room itself though. Though I do believe it is wise to set
and use a specific SPL listening level (on average) when looking for these kind of preferences.
I agree. Subjective listening preferences can be all over the map depending on hearing deficit, conditioning and SPL level. I assume that's a similar curve Harmon published based on their own research? I wish they stated at what average SPL this curve is supposed to be set at and with what type of music. Those two can make a big difference.
The declining curve you show seams to be widely accepted based on research. That slight bump at 1k is part of the fletcher munson curve as well as the slight treble boost and considerable LF boost. HF peaking isn't pleasant to my ears at medium volume levels on up, but I know its usually there to enhance detail up to a certain point. We all know how much bass and treble is typically pushed these days, as well as the compression used to the point of oversaturating the mix (which changes the measured rms level and crest factor).
The declining curve you show seams to be widely accepted based on research. That slight bump at 1k is part of the fletcher munson curve as well as the slight treble boost and considerable LF boost. HF peaking isn't pleasant to my ears at medium volume levels on up, but I know its usually there to enhance detail up to a certain point. We all know how much bass and treble is typically pushed these days, as well as the compression used to the point of oversaturating the mix (which changes the measured rms level and crest factor).
There is no (theoretical..) reason why the reproduction by the end user would need any boost - even a horrible boost, like what "untrained listeners" appear to prefer - of low frequencies.
While preferences can certainly be all over the place, arguing FOR wild deviations from 'flat' response with claims of them being justified by "research" (as in the above graph) is seriously misguided.
Because any capable sound engineer (here I am impyling he/she is competent enough to only use 'flat' operating gear - as well as reasonably 'flat' operating rooms.. - for ALL reproduction! purposes) would already have "adjusted" the spectral balance of the recording in accordance with what the audience he/she is aiming at probably likes in terms of 'sound'.
Hey, I'm only presenting a graph from the research of Toole and company. My preferred curve from tbat set, as I need to draw my own with arrays, is the dashed line (trained listeners). I landed on that curve before ever seeing that graph. Fact is, what measures flat anechoic, would not measure flat, steady state, inside a room at listening distance.
I would not want to listen to that "untrained listeners' curve either. Or it would have to be at very low SPL levels.
I would not want to listen to that "untrained listeners' curve either. Or it would have to be at very low SPL levels.
Last edited:
I think it really depends on what music you listen to and how loud you listen - that is a huge factor IMO. Alot of pop and rock is hyped in mids and treble, plus oversaturated low end, often mixed to cut through on a small portable speaker or car stereo. Its even worse now with MP3s being the popular format these days. I'm all for getting the same vibe of what the engineer or producer intended when it was mastered, but most of the time newer stuff is done by inexperienced people who just want a loud mix that cuts through everything else on YouTube and Spotify. If you're like me, I mainly listen to very well produced music mastered by the top pros ie Doug Sax, Bob Ludwig, Bernie Grundman, etc - these guys know their audience and care alot more.
I know alot of guys aim for ruler flat response when designing a speaker, but flat rarely sounds good with most music, except for maybe jazz, classical. I always design a slight amount of BBC dip (maybe 1 or 2 dB at most) into my stuff, mainly because I usually run everything straight up without any EQ and like to listen on the louder side (requiring more mid dip to keep ears happy). If you're striving for a flat response at any cost, its important to factor in what kind of music you mainly listen to. Yes, an ideal speaker should be able to play anything well, but it doesn't work that way in reality.
Look at this loudness graph and see how non-linear your ears are at various volume levels - thats why its important to compensate FR for listening level. Keep in mind this isn't the exact curve I follow, but it shows the areas where hearing is non linear and something that sounds flat at 70 dB won't do so at 90 dB.
I know alot of guys aim for ruler flat response when designing a speaker, but flat rarely sounds good with most music, except for maybe jazz, classical. I always design a slight amount of BBC dip (maybe 1 or 2 dB at most) into my stuff, mainly because I usually run everything straight up without any EQ and like to listen on the louder side (requiring more mid dip to keep ears happy). If you're striving for a flat response at any cost, its important to factor in what kind of music you mainly listen to. Yes, an ideal speaker should be able to play anything well, but it doesn't work that way in reality.
Look at this loudness graph and see how non-linear your ears are at various volume levels - thats why its important to compensate FR for listening level. Keep in mind this isn't the exact curve I follow, but it shows the areas where hearing is non linear and something that sounds flat at 70 dB won't do so at 90 dB.
Attachments
Last edited:
To the first part, I do agree, and this is all well known and easily demonstrated, that the most neutral sound is when the playback and recording are at the same SPL. But this problem is not one that the speakers can correct for and there is only one playback SPL that mimics the original so there is no "general" solution.
To your second part, this is all conjecture and not at all aligned with my thinking on the subject. Except to say that diffractions from the speaker and nearby room articles is a serious issue whose perception is increasing with increasing SPL. (But these are linear.) There is no driver nonlinearity that enters into the problem unless the system is overdriven. WHen properly designed and setup, the speaker can be quite transparent and non-fatiguing.
Last edited:
There are scientific reasons for a lift at the LFs and a cut at the HFs.
We judge perception level by two things, Pressure level and duration. At LF the sound is so heavily damped (in small rooms) that the duration is cut way down lowering our perception. There is also the F-M curve effect, but I sus[pect that the former is dominate. As to the HF, it depends on the Q of the source. Higher Q - when pointed at the listener - will sound flat when it is flat - there is no room addition effect. For a lower Q systems, the room adds a lot of level and as such needs a subtle rolloff at the high end when compared to the above.
We judge perception level by two things, Pressure level and duration. At LF the sound is so heavily damped (in small rooms) that the duration is cut way down lowering our perception. There is also the F-M curve effect, but I sus[pect that the former is dominate. As to the HF, it depends on the Q of the source. Higher Q - when pointed at the listener - will sound flat when it is flat - there is no room addition effect. For a lower Q systems, the room adds a lot of level and as such needs a subtle rolloff at the high end when compared to the above.