This is a deep topic and its impossible to give a truly meaningful reply, but I will give one non the less. Our current technology in how we record or capture a sound event to how it is reproduced is to limiting to mimic reality. The solutions and technology to reproduce things accurately simply does not exist yet.
But, some loudspeakers and drivers come close, close in such a way that you can sense and feel a holographic sound stage in front of you. It is artificial but at least closer to reality than most. For now, we must be happy that we can get close. Since I have yet to reach a level where I can produce such loudspeakers, I can't for sure say exactly what creates it, but exceptionally drivers are key.
Can't disagree, but some might take that to imply that those recordings might somday reproduce the sound event. What we have recorded on a stereo track is just two spot samples -- not a sound field. Making it sound like a sound field, is, as you say, artificial. It can be very convincingy, but simply having a flat frequency response from those two spots can't do it alone.
I totally agree. I am not going to pretend I know how to record and reproduce a sound field - need wording 🙂 - since I am not a recording artist, but I'll bet you they have ideas about it.
Stereo does if done correctly can in some few cases be played back with very high accuracy, but its limited to what is happening right in front of you, so you lose details about the rest of the room or space. Anyway, we are drifting far from the topic here... lol
Actually the science is mostly settled: you want flattish design axis response, freedom from resonances, and off axis response that matches the design axis response. And smooth clean deep bass, which is about a third of loudspeaker perception!
The major open questions are how wide directivity should be, and how low to control it. The audiophile questions are mostly marketing.
Just a personal tought.
Why does speakers have to sound flat all the time? Hearing is subjective in a large part, and for me flat speakers sound not very good. I've used a lot so called flat high end speakers as studio monitors in (radio and recording) studio's where i did things, and they sound lifeless and fatiguing to me.
I prefer a coloured speaker like the fullrange speakers i have now as main set for listening music. For studio work, yes you need flat speakers to know what you are doing. But not for enjoying music absolutly not for me. I like the coloured sound of vinyl, tube amps and fullrange or coloured speaker drivers.
But each to his own offcourse...
Why does speakers have to sound flat all the time? Hearing is subjective in a large part, and for me flat speakers sound not very good. I've used a lot so called flat high end speakers as studio monitors in (radio and recording) studio's where i did things, and they sound lifeless and fatiguing to me.
I prefer a coloured speaker like the fullrange speakers i have now as main set for listening music. For studio work, yes you need flat speakers to know what you are doing. But not for enjoying music absolutly not for me. I like the coloured sound of vinyl, tube amps and fullrange or coloured speaker drivers.
But each to his own offcourse...
To each his own, but then we can't talk about fidelity of reproduction (which is the name of the game for me) but personal preferences, with what i'm totally fine. As for myself, i do want to hear what the guy in the studio wanted to achieve - or as close as i can come to it.
Floyd Toole said it quite well here:
YouTube
Floyd Toole said it quite well here:
YouTube
Last edited:
Sure, about as settled as how to make an automobile engine: just mix one part gasoline with 14.7 parts air.
Just picking up from my post #4 (which I hoped might raise the thread above trivial arguments about 2 versus 3 dB) and bwaslo's smart insights in post #14 which shows we need to know much more than just "14.7".
It takes much more to produce desirable listening experience than a swell speaker FR (plus polar, plus group delay, plus all the things that Pallas' Science seems to know about speakers will complete the list as of 7 AM as I write this): there is still the room and the not-yet-parameterized ways in which the sound is influenced by all these factors interacting with one another and the listener's circumstances.
B.
No one is arguing against you .. 😀 cone material, drivers size, crossover point, order, choice of capacitor, inductor etc all impact the sound AND it is here we venture into subjectivity. The ruler flat is objective, and the last stop before you have to discover things on your own.
Last edited:
I think you are having a bit of trouble distinguishing the concept of "speaker" from "driver". My point is that with a perfect speaker, you are still just starting on the road to satisfactory results at your chair. Which is why you need the ability to adjust the sound later on.
You also have weaponized the term "subjectivity". There are things that humans best judge and yes, that is "subjective". With a good model of human hearing, maybe a mic could do it. Granted, that is the "long enough lever" fantasy. But that is far from dismissing the term "subjective" in the pejorative sense of just a trivial matter of taste.
B.
Last edited:
Generally, a speaker consist of cabinet, damping material, driver unit (sound producing component) and crossover network.
I have weaponized subjective a bit and given it a slight negative charge, simply because due to our individual genetic code together with a unique footprint of pshycoacoustics render a too large variation in terms of what is considered good and bad. This in turn produce groups and teams who rather often end up "fighting" over who is right or wrong in a "war" that can never be won. The subjective also produce so called audiofool individuals who can be anyone within the audio community. The same individuals will use "that cannot be proven" to attack anyone who argue against using scientifically proven methods because they will claim that simply because you can't prove or disprove something is not a valid method of saying it is real or not - read as religion. There is one place where most of these individuals gather, its the speaker cable camp and I'll bet you, you'll find both fairy's and unicorn corpses because they where sacrificed for their dust, theirs and blood in order to increase the directionality and signal purity in cables mad of neutronstar material, unobtainable for most mortal people ....
Oops, I think I got a bit carried away there for a moment, sorry about that... Subjectivity. We can either view it as reality, neither good or bad or we can argue that to a degree it poses problems, and which camp you chose is ironically a subjective choice. I know I am my own worst enemy from time to time, so I am trying to use science not science fiction in figuring out what good and bad is considered to be. Neutral or ruler flat is one that is not subjective but allow me to move towards a colorless reproduction of that which once was recorded, but who are we kidding, even the recording artist have subjective settings on the mixer table ... well f**** - there are no final answers and sorry for ranting, better go get my cup of coffee and a bucket of calm the **** down ... 😀
Last edited:
Incorrectly applied objectivity, on the other hand, can be just as delusional. I was reading an article until the author illustrates his/her point by running a square wave through mp3 compression and look at the resultant waveform - then I immediately stopped and realized I had been wasting my time reading.
Many speakers show an astonishingly flat FR, but as Zaph puts it, flat FR is good so long as the designer understands the qualifiers, "when", and "where". "Where" has to do with off-axis and thus power response, and "when" had to do with waterfall, etc. As an example let's say a real world loudspeaker has an otherwise "flat" FR but underneath lies a cone break-up issue. How many dB's down would that be to be considered as acceptable? At what frequencies would that be the most offensive? If there is an ultrasonic break-up would that somehow modulate back to audible frequencies?
That audio reproduction is possible at all is because our ears by nature drop a lot of the input signal so that the brain is easily "fooled". Anything having to do with audibility or psychoacoustics is subjective and only became a science because we are striving for a way to describe or even quantify such subjective experiences.
Many speakers show an astonishingly flat FR, but as Zaph puts it, flat FR is good so long as the designer understands the qualifiers, "when", and "where". "Where" has to do with off-axis and thus power response, and "when" had to do with waterfall, etc. As an example let's say a real world loudspeaker has an otherwise "flat" FR but underneath lies a cone break-up issue. How many dB's down would that be to be considered as acceptable? At what frequencies would that be the most offensive? If there is an ultrasonic break-up would that somehow modulate back to audible frequencies?
That audio reproduction is possible at all is because our ears by nature drop a lot of the input signal so that the brain is easily "fooled". Anything having to do with audibility or psychoacoustics is subjective and only became a science because we are striving for a way to describe or even quantify such subjective experiences.
Last edited:
That is not how to "make an automobile engine." Generally, making an automobile engine involves forging and casting metal parts, forming plastics or composites, etc. and assembling them in a set procedure.
If you're going to use figurative language to support a point, make sure it makes sense first...!
I specifically mentioned polar ("off-axis") response. If group delay is an audible problem in a system, chances are there are 1000 other major defects too...
Obviously the room affects the bass. Anyone who cares about sound quality will use multiple subwoofers and equalization to get the bass right. Above the transition region, for the most part, people hear through the room. There are exceptions, though generally so long as a room is comfortable for conversation it is not such an exception. Read Toole and the references therein.
The approach I take is to avoid measurable problems in the first place. If you try to operate a driver too close to non-linear distortion or a cone breakup issue then you aren't helping yourself and will probably find that while the finished speaker measures flat, it doesn't sound subjectively good.
My ears tell me that -40dB (1%) 3rd order and -50dB (0.3%) 4th order and -60dB (0.1%) 5th order non-linear distortion is plainly audible. If I attenuate cone breakup such that it is more than 25dB below the system level then I can't hear any ill effects.
With some rules of thumb like those you can avoid designing something that is going to have audible distortion issues and designing for flat response should not be far off what sounds subjectively best.
Dead flat frequency response will sound good as long as there are no other linear or non-linear distortion issues. If there is a ringing/decay issue or audible non-linear distortion then you will have to deviate from flat frequency response to strike a subjective tonal balance. Too often I see people designing multi-way speakers pushing drivers way too high or low in frequency where distortion is obvious. Then they are either grossly unhappy with the result, or end up with a funky frequency response that they obtained subjectively and don't understand why flat didn't sound good.
The only other reason why you might want to deviate from flat is the Fletcher-Munson curve. For moderate to high listening levels, flat response is correct. For lower listening levels a bass boost and recession around 3kHz achieves subjectively comparable response to higher listening levels. The 'loudness' function on many amplifiers does this, so you may not want to bake it in to your crossover design otherwise when you play the speakers at higher levels it'll be too bassy and have a recession at 3k.
My ears tell me that -40dB (1%) 3rd order and -50dB (0.3%) 4th order and -60dB (0.1%) 5th order non-linear distortion is plainly audible. If I attenuate cone breakup such that it is more than 25dB below the system level then I can't hear any ill effects.
With some rules of thumb like those you can avoid designing something that is going to have audible distortion issues and designing for flat response should not be far off what sounds subjectively best.
Dead flat frequency response will sound good as long as there are no other linear or non-linear distortion issues. If there is a ringing/decay issue or audible non-linear distortion then you will have to deviate from flat frequency response to strike a subjective tonal balance. Too often I see people designing multi-way speakers pushing drivers way too high or low in frequency where distortion is obvious. Then they are either grossly unhappy with the result, or end up with a funky frequency response that they obtained subjectively and don't understand why flat didn't sound good.
The only other reason why you might want to deviate from flat is the Fletcher-Munson curve. For moderate to high listening levels, flat response is correct. For lower listening levels a bass boost and recession around 3kHz achieves subjectively comparable response to higher listening levels. The 'loudness' function on many amplifiers does this, so you may not want to bake it in to your crossover design otherwise when you play the speakers at higher levels it'll be too bassy and have a recession at 3k.
Last edited:
My thoughts are that speakers should have as flat as possible mag and phase response on-axis, and have as much constant directivity off-axis as possible.
Those two goals are inevitably at odds with each other, given the inability to co-locate acoustic sources.....
In my mind, the reality of fitting the best compromise mag and phase curves for both on and off-axis, sets the achievable minimum +/- dB design goal.
And the +/- dB should be no greater than this minimum, reiterating minimum 'built around flat'.
I n terms of the typical on-axis freq response, I like to shoot for less than +/- 1dB with 1/12th octave smoothing, as long as the acoustic design allows for off-axis to stay with +/- 2 to 3, same smoothing. (this is outdoors)
I think the above is true for any speaker,....
.....in any room or venue, for any personal preferences, and for any source material vagaries.
Because I think we make it impossible to sort out what's causing what, when we bring in room/venue acoustics, personal preferences, and source material vagaries into speaker design.
IMO, each of these is ultimately at least as important as speaker response, but each needs to be addressed separately to grow our ability to find what pleases us, and provide repeatable, transferable ability to incorporate different speakers into new rooms etc.
I am a firm believer in using active FIR dsp for linear-phase speaker tuning as described.
And almost just as strong a disbeliever in NOT using the dsp to integrate the speakers into the room, because by far and away, most of the integration needs an acoustic solution, not an electrical solution.
IME, achieving a smooth spectral RT60 through acoustic absorption will drastically and positively affect the tonal balance and clarity of any speaker put into the room.
Speaker position, reducing early reflections, adding diffusion.....all need to be done before any room EQ or dsp processing.
Only then can judicious room EQ be used, most often just to tame bass modes (if room RT60 is done right).
Get all that done, and then it's finally time to address tonal preferences, and source material vagaries.
I try to keep these EQs separate from speaker tuning...I mean, what the heck do they have to do with the speaker ?
They are about what I like and how messed up recordings are 😀
Those two goals are inevitably at odds with each other, given the inability to co-locate acoustic sources.....
In my mind, the reality of fitting the best compromise mag and phase curves for both on and off-axis, sets the achievable minimum +/- dB design goal.
And the +/- dB should be no greater than this minimum, reiterating minimum 'built around flat'.
I n terms of the typical on-axis freq response, I like to shoot for less than +/- 1dB with 1/12th octave smoothing, as long as the acoustic design allows for off-axis to stay with +/- 2 to 3, same smoothing. (this is outdoors)
I think the above is true for any speaker,....
.....in any room or venue, for any personal preferences, and for any source material vagaries.
Because I think we make it impossible to sort out what's causing what, when we bring in room/venue acoustics, personal preferences, and source material vagaries into speaker design.
IMO, each of these is ultimately at least as important as speaker response, but each needs to be addressed separately to grow our ability to find what pleases us, and provide repeatable, transferable ability to incorporate different speakers into new rooms etc.
I am a firm believer in using active FIR dsp for linear-phase speaker tuning as described.
And almost just as strong a disbeliever in NOT using the dsp to integrate the speakers into the room, because by far and away, most of the integration needs an acoustic solution, not an electrical solution.
IME, achieving a smooth spectral RT60 through acoustic absorption will drastically and positively affect the tonal balance and clarity of any speaker put into the room.
Speaker position, reducing early reflections, adding diffusion.....all need to be done before any room EQ or dsp processing.
Only then can judicious room EQ be used, most often just to tame bass modes (if room RT60 is done right).
Get all that done, and then it's finally time to address tonal preferences, and source material vagaries.
I try to keep these EQs separate from speaker tuning...I mean, what the heck do they have to do with the speaker ?
They are about what I like and how messed up recordings are 😀
- Status
- Not open for further replies.