Yes, you're right Z. We can imaging many things, which is why the variation in opinions and tastes in this great hobby exists. 😀
And I'm not even too worried about the intent. Mastering engineers are not try to please ME. They are trying to get the recording to sound its best on a wide variety of systems. And please the artist, and the artist's buddies and newest girlfriend, the record company, etc. ("You don't do heavy metal in Dubly, you know.")
The target audience is NOT us. Mastering for SACD and DVD-A is much more aimed at folks like us and you can usually hear the difference.
But we're getting OT here.
The target audience is NOT us. Mastering for SACD and DVD-A is much more aimed at folks like us and you can usually hear the difference.
But we're getting OT here.
Sometimes mastering engineers try to please themselves as well, although it rarely happens these days. I've heard both really awful over-compressed stuff, and really great things coming from the one and same mastering engineer. The more a record is expected to sell, the less freedom is given to the mastering engineer I guess.... And isn't it funny that records sounded much better in the days before the word "remaster" was invented... ?
Once upon a time, some brave people, despite bad reputation since the days of Olson, decided to give OB a try. They liked what they heard. Then, as if conducting the world's dumbest most confused experiment, tried to figure out which single aspect - of the endless variables and near-infinite room settings - was THE factor.
Although it makes little conceptual sense for an engineer or manufacturer, for DIY experimenters, an approach that often works well is spreading around the factors and splitting the differences, and so on. OB concepts do that nicely by pumping some sound forward, some back, some sideways, some resonances... but not too many bad ones.
Like with ESLs in its ways, OB sound is so tasty in its ways, you can even get away with wholly radical (read: shoddy) notions like a single cone driver for great parts of the frequency compass, missing fundamental tones, etc.
One of the few solid generalizations I have drawn from much reading at this forum is that a simulation curve is a weak predictor of microphone response which, in turn, isn't helpful except as a first crude target. My further belief, somewhat paradoxically, is that there is fair agreement on what sounds good in a specific setting and sources, as Toole says.
Kind of random luck if a speaker sounds pretty good on first installing it, whatever it is. You've to to tweak it on your favorite music, as nearly everybody agrees.
Now back to the detailed analysis of the world's dumbest experiment.....
Hint: if you want to conduct an interpretable experiment, you need to remember the "Rule of One Variable" where you change just one thing at a time.
Although it makes little conceptual sense for an engineer or manufacturer, for DIY experimenters, an approach that often works well is spreading around the factors and splitting the differences, and so on. OB concepts do that nicely by pumping some sound forward, some back, some sideways, some resonances... but not too many bad ones.
Like with ESLs in its ways, OB sound is so tasty in its ways, you can even get away with wholly radical (read: shoddy) notions like a single cone driver for great parts of the frequency compass, missing fundamental tones, etc.
One of the few solid generalizations I have drawn from much reading at this forum is that a simulation curve is a weak predictor of microphone response which, in turn, isn't helpful except as a first crude target. My further belief, somewhat paradoxically, is that there is fair agreement on what sounds good in a specific setting and sources, as Toole says.
Kind of random luck if a speaker sounds pretty good on first installing it, whatever it is. You've to to tweak it on your favorite music, as nearly everybody agrees.
Now back to the detailed analysis of the world's dumbest experiment.....
Hint: if you want to conduct an interpretable experiment, you need to remember the "Rule of One Variable" where you change just one thing at a time.
Last edited:
I would fairly strongly disagree. Way back in the dark ages, I used Xopt for crossover development, and Leap (1???) for box design. I have been using Leap 5 for 9-10 years (?) for enclosure and crossover design, and in both cases, especially with Leap, the simulation output is very very close to mic measurements (MLSSA), and, subject to the considerations below, matches the perceived audible response.
There are some valid reasons why speaker builders might feel that there is a disconnect between measured and perceived results, but these are causes that can be dealt with.
Consistency in mic placement is critical; we used a measuring tape each time to set the mic-baffle distance, and microphone height (eqidistant between mid and tweet). And we used an alignment rod to be sure the microphone was pointing exactly at the mid-tweet axis.
I feel that spectral decay (waterfall) measurements are very important, as the audible effects of ringing are much greater than the frequency response alone would imply. Since the length of time of the ringing is so long, the power of the ringnote is disproportionately great. As a result, woofer cone breakup, if, say, 20 dB down at the breakup frequency, will be easily audible, although it will barely, if at all, show up on the composite frequency response. I trap the breakup region, and that, plus aiming for 35 dB or so down for the remnants, prevents the breakup remnants from harshening the upper mid.
For some reason, I tend to notice diffraction effects disproportionately, and a well rounded, (and with exterior felt) cabinet reduces those markedly. Also, reflections from the back of the cabinet interior will exit back through the cone, as it is acoustically transparent; again the audibility is out of proportion to the measured effect, because of the 2? ms or so delay. These are reasons why testing should be conducted in a proposed "final" cabinet design, fully stuffed and positioned. This applies to the initial driver response curves that are the starting point for the simulator. Those initial curves must be taken with the microphone in the standardized position as well.
Obviously crossover component tolerances can play a part, especially for relatively high Q traps. If the value in the circuit isn't what the simulator thinks it is...
My listening preferences tend towards acoustic instruments, symphonic, jazz and blues, '50's and 60's RnB, and relatively little contemporary pop. Had symphony season tickets for many years. I dislike peaks and false dynamics, hence my outlook on some of these issues.
Yes indeed, reduce diffraction as much as you can.
I'm sure you realize this, but if a cone was acoustically transparent, it could not make noise. But yes, I agree with you--fill your enclosures as best you can with sound absorbing material and reduce break up as much as possible.
Dan
The reflected wave moves the cone, and is recreated on the front, with remarkably little loss. It's easily seen with an impulse response, with an unstuffed, unpadded cabinet, with a vertical back wall. Since the reflection is not significantly delayed, only moderately attenuated, and the signal is "repeatered" with some fidelity, I'm not sure what else to call the phenomenon that conveys the issue as clearly.
Very interesting discussion here with many take-away insights.
1. As with any human effort outside the screwball world of people who think of themselves as "Artists," there's some control, feedback, boss, market-discipline, etc. that "controls" people like recording engineers. So what he/she produces goes off to somebody called the Producer or the Artist who makes a final call. I don't know much about that business, but you get the idea.
2. "Am I reproducing Carnegie Hall in my living room" or the derivative version of panomaniac ("bread lover"?... whose thoughts I often admire) offers, "well... at least the timbre resembles real sound...." is the deep debate that never goes away. This morning, all I'd like to add relates to a variant of the "down the hall test." Have you ever in your whole life, except when in an altered state of consciousness, really thought somebody was playing a real piano in your music room, even from down the hall? Even just a flute? In as much as the answer is no, could it be that nobody is really trying to create recordings that do that? Or at least in the sense that the binaural earphone fans really truly honestly hope they are.
My question in the previous paragraph is logically separate from the question of whether or not a loudspeaker should (in some sense) reproduce the recording "accurately."
1. As with any human effort outside the screwball world of people who think of themselves as "Artists," there's some control, feedback, boss, market-discipline, etc. that "controls" people like recording engineers. So what he/she produces goes off to somebody called the Producer or the Artist who makes a final call. I don't know much about that business, but you get the idea.
2. "Am I reproducing Carnegie Hall in my living room" or the derivative version of panomaniac ("bread lover"?... whose thoughts I often admire) offers, "well... at least the timbre resembles real sound...." is the deep debate that never goes away. This morning, all I'd like to add relates to a variant of the "down the hall test." Have you ever in your whole life, except when in an altered state of consciousness, really thought somebody was playing a real piano in your music room, even from down the hall? Even just a flute? In as much as the answer is no, could it be that nobody is really trying to create recordings that do that? Or at least in the sense that the binaural earphone fans really truly honestly hope they are.
My question in the previous paragraph is logically separate from the question of whether or not a loudspeaker should (in some sense) reproduce the recording "accurately."
Last edited:
Let me say right off that I am not against the scientific advancement of sound reproduction.
From the quote, I see there is a basic difference in our views. Yes, you can successfully "engineer" a system to perfectly meet some narrow physical criterion ("SPL at 1 meter at 100 Hz"). Some of these precise measurements relate to beautifying parameters that look awful when the visual chart is eyeballed but may have doubtful relevance to hearing nice music.*
By "mic measurements" I meant at listener's ears. And that in turn, still has a long ways to go to be "the sound the listener likes."
*Adherents of each enclosure faith (I believe in ESL With Bass Horn), have widely different notions of what design issues matter. Some ignore issues of diffraction or beaming or low bass, etc.
I think you need some time using a measurement system and crossover simulator software. You are completely missing his point about the importance of proper set-up and how it relates to good repeatable measurements. Good simulators are just like the old computer expression "Garbage in Garbage out". The quality of the predicted responses are proportional to the care taken during the measurement process. It's not some "narrow physical criteria" A bad mmeasurment set will give you a bad end result.
Rob🙂
Curmudgen, maybe acoustically translucent?😕 🙂 Check out the graph of 2 of the same model pro speaker in the same size box, one with fiberglass filling, one without:
In the worst case scenario, you won't need the impulse to see its effects.
Dan
In the worst case scenario, you won't need the impulse to see its effects.
Dan
@ dtm: Glurk! 😱
@ Robh3606: Wish I'd said it that well.
@ bentoronto: The tool is not the end, but the means for achieving the desired end. If you look at my previous post, you'll see that the response is not flat; it does however match quite closely the initial target that I set up in the simulator.
If you hear something you'd like to change; say, male vocal chestiness, it helps a great deal to know the shape of the bump; size, frequency, Q if appropriate, etc. If you try to modify the crossover by ear alone, you need to possess a rare set of skills to succeed consistently. For most, simulation and measurement tools help keep one out of the quicksand.
The tool doesn't "know" or care what you hear, or what you decide to do about it. It just helps you do what you feel needs to be done. The act of measurement does not require you, or commit you, to design for, perhaps, a ruler flat response at 1m. However, it does allow (assist) you to do so, should you so wish.
This hobby (like photography too, I think) straddles the ever-restless intersection of the technical and the artistic. I believe that some appreciation and understanding of both spheres is required for consistently satisfying results.
@ Robh3606: Wish I'd said it that well.
@ bentoronto: The tool is not the end, but the means for achieving the desired end. If you look at my previous post, you'll see that the response is not flat; it does however match quite closely the initial target that I set up in the simulator.
If you hear something you'd like to change; say, male vocal chestiness, it helps a great deal to know the shape of the bump; size, frequency, Q if appropriate, etc. If you try to modify the crossover by ear alone, you need to possess a rare set of skills to succeed consistently. For most, simulation and measurement tools help keep one out of the quicksand.
The tool doesn't "know" or care what you hear, or what you decide to do about it. It just helps you do what you feel needs to be done. The act of measurement does not require you, or commit you, to design for, perhaps, a ruler flat response at 1m. However, it does allow (assist) you to do so, should you so wish.
This hobby (like photography too, I think) straddles the ever-restless intersection of the technical and the artistic. I believe that some appreciation and understanding of both spheres is required for consistently satisfying results.
Curmugeon -
Yes, perhaps I am Rip Van Winkel, with my impressions fixed in place 40 years ago.
Perhaps you could post, say 3 charts for different speakers or rooms you've made in which I could see a simulation trace and an ear-level mic measurement(s) on the same axes for trustworthy comparison of how well the simulation anticipates the outcome. Hope you won't fall back on those easy-to-do up-close mic traces, voice coil impedance curves, or hide the fine-grain with third-8ave smoothing.
Since you have strongly ridiculed me for lacking your productive experiences, I assume you've mastered that kind of procedure any number of times. And, it should be piece-of-cake for you to draw three from your extensive experience to show us how helpful these simulations are.
Thanks. While I have my expectations, I will be very happy to learn techniques I can use myself.
Yes, perhaps I am Rip Van Winkel, with my impressions fixed in place 40 years ago.
Perhaps you could post, say 3 charts for different speakers or rooms you've made in which I could see a simulation trace and an ear-level mic measurement(s) on the same axes for trustworthy comparison of how well the simulation anticipates the outcome. Hope you won't fall back on those easy-to-do up-close mic traces, voice coil impedance curves, or hide the fine-grain with third-8ave smoothing.
Since you have strongly ridiculed me for lacking your productive experiences, I assume you've mastered that kind of procedure any number of times. And, it should be piece-of-cake for you to draw three from your extensive experience to show us how helpful these simulations are.
Thanks. While I have my expectations, I will be very happy to learn techniques I can use myself.
Last edited:
I'm very sorry that you felt that I ridiculed you; I had no intention whatsoever to do so and apologize profusely.
And your implication is correct; I never take measurements at my listening position(s). Room reflections and the differences you get from a few inches difference in mic position seem to me to make such an exercise pointless. You just get into a morass; should I be seated in the chair, leaning forward, or settled back, where does the mic go so that my head does not change the soundfield?
The brain is capable of sorting out the mess; and that, in the end is the final determinant. Careful measurement, clean, up close measurement, gives a view of what the sound started out as; after that, subjective issues become important, such as the effect of room treatment on what you hear.
We are nowhere near as far apart as you seem to feel.
And your implication is correct; I never take measurements at my listening position(s). Room reflections and the differences you get from a few inches difference in mic position seem to me to make such an exercise pointless. You just get into a morass; should I be seated in the chair, leaning forward, or settled back, where does the mic go so that my head does not change the soundfield?
The brain is capable of sorting out the mess; and that, in the end is the final determinant. Careful measurement, clean, up close measurement, gives a view of what the sound started out as; after that, subjective issues become important, such as the effect of room treatment on what you hear.
We are nowhere near as far apart as you seem to feel.
Isn't that a hoot. Just last week, I was in exactly the same quandary as you. I sat in my sweet spot chair thinking that's important to have constant. Therefore I had to have the mic on a tripod 6 inches behind my bald spot and two inches above. Damn, if half the time the meter didn't go up when my perception went down and vice versa (OK.. maybe that adds up to 100% of the time, but that is what it felt like). When I tweaked my subwoofer EQ by crude mic indications, it sounded better. Go figure.
About the "brain sorting things out" - well, that's no answer at all. But explaining just what the brain does and doesn't do is crucial and it is the giant hole in your argument about working by simulations because they have no model of hearing. It is fair to say that hearing is part of the loop, eh.
Clearly, the brain does "learn" something about rooms and thereafter re-interprets sounds in light of its model of the room. You may never have heard it put that way, I think it is fair picture.
On the other hand, back to the thread (since it may take a lot time to see those traces I asked about), the brain certainly is not sorting-out loudness. I ask any amateur engineer out there, how come? Seems a trivial correction for a computer to do, so why not the brain? (I think I have an answer.)
Last edited:
When you are doing measurements for a crossover simulator you pick a point in space, typically 1-2 meters on the tweeters axis. You don't do the measurements to far away or you will have to much interference from the room even in a gated measurement. Here are a couple of compression driver compensations done in LEAP using Clio to measure. The actual and predicted curves are overlaid for each driver in the pair.
Rob🙂
Rob🙂
Attachments
Last edited:
They are not supposed too. They are there to predict a acoustic response based on the provided measurement curves and the as designed crossover. Remember you are trying to simulate anechoic measurements not the room response plus the speaker response. What you end up with are FR curves Imp curves and so on. All this without the room influence.
Rob🙂
- Status
- Not open for further replies.