is this an admission that there is an audible problem with oblate spheroid waveguides?
this is not meant as any sort of an insult but if your being honest and admitting there is a flaw i tip my hat in awe and admiration to a truly great man.
You could spend your whole life doing tests like this, only to find that your hearing has a dip in the 1khz region and so all your designs were perfect....but only for you! Imagine the power of bias you would have to overcome to accept something like that!!
I think we could look at this from another angle. Designing speakers is a lot of work, and a huge amount of unknown.
Perhaps what we need is simply a single reference point that is common across designers.
Like the sound mastering studios and DJs of the 80s did when they all bought Yamaha NS10 monitors, we need an agreed small speaker design that meets all of the measurable, and most importantly provable current parameters known to affect sound quality to a good level.
No design criteria should be included in it that does not have a solid basis in an ABX trial, but outside of that, it should be as cheap and as easy to build as possible.
Why only provable design criteria? because once you let in other criteria, you cannot justify why you let 1 slip by, but not another.
Everyone can build a set and:
-Use it to refer back to for their own designs
-Use to explain how other creations sound in comparison.
-Commercial design evaluation
-Testing of different modifications (perhaps the design could lend itself to be adaptable to different box designs like TL versions, but without changing the crossovers or drivers)
The real benefit would actually be that we as a community would use the 'hive mind' to cancel out many biased decisions and impressions for each other. For example, you are certain that your speaker sounds better with a crossover set at 1.6khz but it doesn't measure as well? we can all try it and see, does this subjective impression hold true across 50 or 100 people?
If yes, there may be more too it. We investigate further, divide into 2 camps, those that try to prove it, and those who try to disprove.
If no, your bias (or room) got the better of you.
On that note, I would suggest that tests are done outside in a open space for an apples vs apples comparison.
Each person should also undergo a basic hearing test before adding their data.
I would expect multi-variant correlation analysis to begin to show answers that we had not thought of before, as long as the data is reliable, and enough people are involved we would start geting some answers.
This way we could work our way through the list of tweaks and criteria that many experienced designers hold as true and test them without the ball ache of getting hundreds of people in a room for a proper ABX test, which of course is NEVER done correctly unless every person in that room sat in the sweet spot (just for starters!)
@lbstyling, do you know what Dr. Geddes wife does for a living? 
I see the above idea as seriously flawed, unless you'd also include a listening room to go with the set of reference speakers. Even though you mentioned the room (only once), this part is one of the most important pieces of the puzzle.
I see no use in a reference set of speakers without defining the way it should be setup too. And that's just not going to work i.m.h.o.
I see the above idea as seriously flawed, unless you'd also include a listening room to go with the set of reference speakers. Even though you mentioned the room (only once), this part is one of the most important pieces of the puzzle.
I see no use in a reference set of speakers without defining the way it should be setup too. And that's just not going to work i.m.h.o.
On the other hand, you would have that same dip in your hearing when listening to live singing and musical instruments.
I don't think an NS10 would be considered anywhere near a "reference" as the word might be used in this thread! The reason it was popular was because its sound translated well to "mid-fi" home and car audio systems, through which most popular music was listened to.
I suggested testing outside in an open space.
a deep null (for instance) would mean you are less sensitive to say having a time error at this frequency, or distortion, which may mean a design choice sounds good to you, but not to others, so no.
The NS10 is an example of what another industry uses as a reference point. I wasn't suggesting it is a reference level speaker. Just a reference point. I suggested building a new design.
Last edited:
@lbstyling, do you know what Dr. Geddes wife does for a living?
I see the above idea as seriously flawed, unless you'd also include a listening room to go with the set of reference speakers. Even though you mentioned the room (only once), this part is one of the most important pieces of the puzzle.
I see no use in a reference set of speakers without defining the way it should be setup too. And that's just not going to work i.m.h.o.
See my previous post for comments on listening rooms.
When you say 'defining the way it should be setup' I'm assuming you might mean electronics? We could specify the requirements for an amplifier, and even suggest the cheapest one to meet the specifications. A class D amp could come to as little as $60 or $80 if the listening level is kept low enough to avoid clipping.
I meant the room, as most of us want to listen inside and not all of us can drag the stuff outside to listen to it.
Speaking only for myself I opted for speakers that could have a shot at performing great in my room under my specific circumstances. We will always have to face this "room" problem at some point.
You speak of relatively small speakers, I know that won't work for me, as I like what that bottom end can add to/aid in our perception.
The next flaw is assuming 'listening levels can be kept low enough.' Perception certainly changes with listening levels. It is hard enough to know what listening levels were used in mixing, we'd have to set a reference for that too for anything to translate, as it will definitely influence our perception.
I listen to an average of ~85 to 87 dB. I know there are lots of people that want louder SPL numbers and lots of people that listen at softer levels too.
We never get to know what levels were used in production.
There are just too much differences between all of us, the room, the preferences, the types of music we play and even our budget that a reference is quite impossible to determine.
It isn't coincidence that there are many different preferences from people in our audio circle. Our listening habits are probably as different as our taste in music is.
We will probably never be able to settle on one concept as a reference. What should it be? We have too many preferences:
Full range single driver
Full range with bass support
Full range with a tweeter
2 way
3 way
4 way
Horns
Planars
Arrays
Sealed
Ported
Open baffle
Naked drivers
etc. etc. This list will get too long as the possibilities are endless. Each category can be subdivided. Once someone gets to a point he/she enjoys, it will be defended like crazy. No other concept will ever come close. The whole world will be declared crazy for not believing it is the only concept that leads to this holy grail.
I'd rather see more people posting what they do get at the listening spot with measurements. But that's not going to happen either.
Many don't believe a measurement can show us how something sounds. I guess many don't even want to know how it measures if they enjoy it as much as they do.
It might even take the "magic" out of it for them.
Probably the biggest variable in this whole game is that brain between the ears and it's influence on perception. Sadly, we cannot standardise that.
Speaking only for myself I opted for speakers that could have a shot at performing great in my room under my specific circumstances. We will always have to face this "room" problem at some point.
You speak of relatively small speakers, I know that won't work for me, as I like what that bottom end can add to/aid in our perception.
The next flaw is assuming 'listening levels can be kept low enough.' Perception certainly changes with listening levels. It is hard enough to know what listening levels were used in mixing, we'd have to set a reference for that too for anything to translate, as it will definitely influence our perception.
I listen to an average of ~85 to 87 dB. I know there are lots of people that want louder SPL numbers and lots of people that listen at softer levels too.
We never get to know what levels were used in production.
There are just too much differences between all of us, the room, the preferences, the types of music we play and even our budget that a reference is quite impossible to determine.
It isn't coincidence that there are many different preferences from people in our audio circle. Our listening habits are probably as different as our taste in music is.
We will probably never be able to settle on one concept as a reference. What should it be? We have too many preferences:
Full range single driver
Full range with bass support
Full range with a tweeter
2 way
3 way
4 way
Horns
Planars
Arrays
Sealed
Ported
Open baffle
Naked drivers
etc. etc. This list will get too long as the possibilities are endless. Each category can be subdivided. Once someone gets to a point he/she enjoys, it will be defended like crazy. No other concept will ever come close. The whole world will be declared crazy for not believing it is the only concept that leads to this holy grail.
I'd rather see more people posting what they do get at the listening spot with measurements. But that's not going to happen either.
Many don't believe a measurement can show us how something sounds. I guess many don't even want to know how it measures if they enjoy it as much as they do.
It might even take the "magic" out of it for them.
Probably the biggest variable in this whole game is that brain between the ears and it's influence on perception. Sadly, we cannot standardise that.
When I build a speaker I hope they'll be ok and expect them to be crap.
The first few times I play them all I hear is flaws.
This only changes after some a/b'ing with other speakers and opinions of friends whose ears I trust. Except when they really are crap...
This describes your expectation, perception, realisation and adaptation.
Actually, everyone's hearing has frequency response anomalies like this that differ from person to person, at least on a biological/mechanical level in the ear.
Everyone has different shaped ears for a start, which obviously affects the frequency response received in the bottom of the ear canal, and then there is the sensing parts of the ear like the cochlea and nerves that differ from person to person and change in effectiveness for a given person as they age, causing a shift in frequency response of the ear itself. (Such as reduced high frequency sensitivity, or specific frequency range losses due to industrial noise exposure)
However there is a tremendous amount of "compensation" applied by the subconscious processing in the brain to "EQ" out most of the frequency response errors present in the physical response of our ears.
Because of this, despite an "anomalous" frequency response of the ear itself, we can perceive a neutral, balanced frequency response when we hear it, with some caveats.
There are two main caveats that can cause our perception of tonal balance to become skewed.
1) Extreme hearing loss at certain frequencies. The brain seems to be able to "EQ" the response to compensate for moderate losses in hearing at certain frequencies, however it can only adapt so far. So if you have moderate hearing loss at some frequencies (by say a few dB) then the brain will adapt over time. And if that loss of hearing occurs over time the loss and adaption will be gradual enough that you don't even notice anything amiss unless it gets really bad - beyond the ability to adapt to the problem.
A good example of this is ear wax build up over many months - gradually your physical high frequency response declines, but the brain is also "turning up" it's sensitivity to high frequencies at the same rate, so you don't necessarily notice.
I used to get a lot of problems with excessive ear wax build up when I was younger (thankfully getting older has largely solved that!) and I'm very familiar with the perceptual effects of this - that you don't really notice it happening as it gets worse but when you get your ears cleaned out suddenly everything is louder and much "crisper" to the point where you think the treble has been turned up uncomfortably high on everything in the world.
And it can take a few days for the new status quo to sound "normal" again. This is the brain adapting in the opposite direction, gradually "turning down" the treble in response to everything seeming too bright. Walking through a noisy environment like a busy city before that re-adaption takes place can actually be uncomfortable.
While the brain can adapt the frequency response for moderate hearing loss it can't do anything about your sensitivity to low volume sounds - so a hearing test that probes your lower threshold of hearing at each frequency will clearly show any frequency dependent hearing loss that you may have that you brain is "hiding" from you at normal volume levels. This can be used to predict how close you are to actually consciously noticing hearing problems.
If hearing loss at certain frequency ranges is excessive then you will no longer be able to perceive a neutral tonal balance as it is out of the range of possible adaption.
2) Excessive exposure to tonally unbalanced sound will eventually "skew" our brains EQ compensation system towards trying to make this unbalanced response sound neutral and normal.
This is a MAJOR problem for anyone designing and tweaking speakers and spending a lot of time (probably too much time) listening to said speakers, especially if we are not well aware of this adaption effect and how it can fool us.
I believe this is the main underlying phenomenon that causes design by subjective listening tests only to "go off the rails" and end up nowhere near neutral, if not also backed up and anchored by good measurements. The more we listen to something unbalanced like an unfinished speaker the more we like it and think it sounds normal. This becomes a circle of confusion in its own right just between you and the speakers.
We're trying to judge the tonal balance of the sound purely by listening, but not recognising the fact that excessive exposure to that imbalance causes it to sound more neutral to us over time because our brain thinks there is something wrong with our ears and tries to adapt to it.
When I see reviewers talk about speakers "sounding better after break-in" I'm dying to tell them "sorry bud, it's not the speakers that are breaking in - it's you!".
If the speakers sound bad at first but sound better after you've listened to them a lot, it's not the speakers that have changed, it's acclimatisation of your brain to their response errors making them seem better.The fact that any acclimatisation period is needed in the first place is typically a sign that the speakers are NOT balanced. (Provided that your perception isn't already skewed by excessive exposure to other bad speakers that are bad in a different way, see below)
Why is there any adaption at all in the brain ? I think the answer is pretty simple - there is nobody at the baby making factory to "calibrate" our ears as we come off the production line, everyone's ears are significantly physically different (just look at how different and incompatible different people's HRTF's can be for binaural recordings) and they also change as we get older, get injured, have noise related hearing loss etc...
So there has to be some means of self calibration - a rough approximation of it would be that the brain has a spectral profile of what a "normal" average frequency distribution of natural sounds are, and does a very slow, (days, weeks) time averaged adjustment of our subconscious EQ to make "most" things that we hear come into line with our expectation of the overall balance of sounds around us.
If there is a persistent error in the response that is there all the time on all sounds, such as the ear wax scenario, gradual adjustments are made to compensate until things fall in line again.
Perhaps it's smarter than this - perhaps more priority is given to familiar recognisable sounds like voices etc. I don't know. All I know is that the adaption does take place, it takes time, and depends on exposure.
So what can we do about it as speaker designers, so we don't go down a rabbit hole of a circle of confusion ? A few things:
1) Use measurements as a starting point and anchor. If a speaker measures really un-flat then believe the measurements, not your ears if you have been listening to these unflat speakers for a long time as you will have adapted to them.
It's OK to deviate slightly from flat when tuning a speaker by ear (probably necessary in fact due to room effects of directivity etc) but if your bass is 6dB lacking because you decided you don't like baffle step compensation, and you think that sounds OK even though the speakers measure 6dB short in bass, well you're wrong...
If you get "lost" when tweaking or redesigning the crossover, always use measurements to get yourself back to a known starting point or head back in the right direction.
2) Don't over expose yourself to a speaker that you know is measuring un-flat - such as one that you haven't finished the crossover for yet or haven't even measured yet. You will adapt over time to its sound and become less aware of the response error of the speaker, which will colour your judgement of how good or natural it sounds.
This is particularly a problem if the speakers under test are your "daily drivers" that you listen to all the time for music, watching TV etc, as is the case for me. Even though I'm well aware of the effect I'm describing it's still hard not to get caught out by it because I listen to them so much.
3) Try to expose yourself to "natural" sounds a lot more than your speakers. This will help prevent your brains adaption from "drifting" towards correcting your speakers. Those that like to use natural, "live" music like concerts as a reference have the right idea. You want to listen to natural sounds as much as possible and potentially coloured sounds like your in development speakers as little as possible to avoid drift of your internal compensation.
4) Mentioned above already, but first impressions are probably the best for a new speaker, especially if you listen to plenty of natural sounds/music and thus your internal balance is not already skewed in a certain direction. If a speaker sounds tonally wrong at first listen it probably is. A great speaker should sound balanced and "right" from the first few seconds that you start to listen to it.
If you have to listen to a speaker for days to weeks to come to appreciate it then it is almost certainly wrong. When it comes to tonal balance at least. For some other flaws like resonances these may not be immediately apparent as they may "trigger" only on certain music so may take longer listening to identify.
I shall now don my fire resistant suit and exit stage left.
Last edited:
Far, far safer for all of us to have doofus arm-chair speculation about how hearing perception works than about how to build a bridge.
B.
I'm assuming that's supposed to be an insult, but I can't actually understand what you're trying to say... thanks for the thoughtful input on the subject.
that's not quite true 85 db is used as a standard for monitoring but don't take my word for it check this out Establishing Project Studio Reference Monitoring Levels |
@turk 182, I do realise that, which is why I do use that 85 to 87 dB average, with a cheap Radioshack SPL meter always nearby.
However we can never be sure about it. It is so easy to turn up the volume with clean sounding speakers that I decided to use a standard level, with (R128 standard based) level compensation active from within my playback chain.
Even though it is widely assumed 85 dB is used, we can never actually be sure about it.
However we can never be sure about it. It is so easy to turn up the volume with clean sounding speakers that I decided to use a standard level, with (R128 standard based) level compensation active from within my playback chain.
Even though it is widely assumed 85 dB is used, we can never actually be sure about it.
DBMandrake
that post creates a Kobayashi Maru scenario
we are adaptive to bad sound but somehow retain the ability to perceive neutral balance?
what's dominant here?
using live concerts as a reference? classical music cannot be recorded without compression and loss of spectral information which is the big beef for purists so how can a fair comparison be made?
worse yet is a popular music lover with original source sounds that contain "synthesis" and amplified sources. how is a rock concert with a mega watt PA in an arena any sort of reference for the "same" in our living room?
if daily listening sources like TV skew our perception we're really screwed.
the one thing i can agree with is if it doesn't sound right at the first listen it's not right but then that suggests trusting one's ears and perception...but most of your post was suggesting we shouldn't so which is it?
that post creates a Kobayashi Maru scenario
we are adaptive to bad sound but somehow retain the ability to perceive neutral balance?
what's dominant here?
using live concerts as a reference? classical music cannot be recorded without compression and loss of spectral information which is the big beef for purists so how can a fair comparison be made?
worse yet is a popular music lover with original source sounds that contain "synthesis" and amplified sources. how is a rock concert with a mega watt PA in an arena any sort of reference for the "same" in our living room?
if daily listening sources like TV skew our perception we're really screwed.
the one thing i can agree with is if it doesn't sound right at the first listen it's not right but then that suggests trusting one's ears and perception...but most of your post was suggesting we shouldn't so which is it?
Both, neither. Unfortunately the truth is somewhere between, I say unfortunately because that's not sexy and doesn't sell well.
Use measurements to guide you? As that would be a way more neutral option than any unbalanced hearing system.
Seems like solid advise. While at it, try and hear the room wherever you go. Your brain will filter it out on auto pilot, but with a bit of practice you can learn to at least hear some of it's effects.
Seems like solid advise. While at it, try and hear the room wherever you go. Your brain will filter it out on auto pilot, but with a bit of practice you can learn to at least hear some of it's effects.
Last edited: