|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
75.71.37.161
Many of us use some sort of real-time sound level meter to position speakers and/or configure equalizers to get smoothest (i.e. flattest) frequency response at the listening position.
Yet a measured flat response at the upper end of the spectrum might sound subjectively way too bright and a measured flat response at the low end might sound subjectively too anemic. And I'm not talking about problems of calibration with the meter itself.
I remember an "ideal" measured response curve published many years ago by Bruel & Kjaer (or somebody)that addressed this.
I'd love to resurrect this topic, especially since digital correction is such a hot item right now.
Follow Ups:
Just fyi, I haven't forgotten about youse guys Throwback, 3db and Dave K.
It's just that I have limited time to re-write and reference what has already been done.
For the moment, look at:
Beranek: "Music, Acoustics & Architecture" (a landmark work)
Knudsen/Harris: "Acoustical Designing In Architecture"
Everest: "Master Handbook of Acoustics".
and report back.
:)
Download the user manual for the Quatros, and read the section on adjusting the low-frequency levels. He has a "user" section, and a section for technicians with calibrated SPL meters and matching test CDs. He warns against setting response for flat levels with the meter.
WW
"A man need merely light the filaments of his receiving set and the world's greatest artists will perform for him." Alfred N. Goldsmith, RCA, 1922
If your system is doing damage to the sound at low levels bringing the levels even lower than you would normally hear in nature. Since the sensitivity with level affects YOUR hearing and not the sounds per se it means you have the same bias with real live sounds.
However, if your stereo is not sensitive to small fluctuations at lower levels (most low sensitivity speaker systems suffer a "drop out" below a certain SPL) then it will sound lacking in both bass and highs compared to "the real thing". This is one of the biggest problems in audio and its affect on realism, IMO. It also manifests itself at higher volumes when a soft sound is trying to be discriminated (my late friend Allen Wright called it DDR, downward dynamic range).
A proper compensation would need to see how much of a deviation is being made and applying that dynamically with each change in level. The compensation would NOT be to the Munson/Fletcher curves but to whatever the systems original target curve would be at say 85db.
The problem is that you are always correcting what has already come before with a feedback system. With high dynamic music you might be applying the right curve correction at the wrong time!! A proper system would need to have the right response measured for a given speaker/amp combo and that file loaded into memory. THEN for the music being played the system would need to analyze the WHOLE recording in advance to be able to apply the right correction at the right time. Of course the listener would have to include their target reference level (min, mean and max SPL levels) or else it would STILL be incorrect with a random volume level selected.
A non-trivial problem IMO but still solvable.
Morricab:
We may be talking about two different things. Or are we? I was talking about the phenomenon that a system that measures flat at the listening position will sound bright to the ear at normal listening levels. If I read Fletcher-Munson right, as one reduces the volume the measured response will remain the same but both the highs and lows will drop off as perceived by the ear.
Tricky stuff, eh what?
Well, in that case you are talking about power response vs. on-axis response in a room. A truly flat measuring speaker will sound bright if the dispersion is even and wide and you have reflective surfaces. In a large room or with a speaker that has quick rolloff in off-axis measurements, it is less likely to sound bright.
If you are listening to a recording at lower than natural volume levels then to get a "live" feeling there would need to be boost because of your changing perception and the fact that it is not at the natural level for that given performance (really only applicable with uncompressed acoustic music). Enter the loudness button, however, the bigger problem is how stereos handle low level signals and as you lower the volume then subtle sounds will simply drop out because the system cannot generate such small responses.
"the bigger problem is how stereos handle low level signals and as you lower the volume then subtle sounds will simply drop out because the system cannot generate such small responses.""because the system cannot generate such small responses.".
Show me. This is easy to prove. Is the signal there, or, is it not there?
I understand that your friend believed it, and that you believe your friend. The emotional factor is huge in situations like this, so it's essential to do our best to eliminate personal bias.
Moving on...
If one plays a recording of an orchestra where the live peaks were 105 dB, and sets the playback volume to, for example, peaks of 80 dB (a reduction of 25 dB), minute details will obviously be lost. This is common, and, to be expected.
Often, it's a result of hearing acuity and/or the ambient noise level in the playback environment.
You'll have to provide some evidence that the signal simply isn't there, and that it's not because the listener couldn't hear it.
Your're a smart and experienced guy, but your argument that the system can't reproduce very low level signals is suspect.
:)
Edits: 06/22/15
It is not if the signal is there or not it is if the driver that is supposed to reproduce it will respond to it or not. Below a certain level there are resistive forces (mostly mechanical) that will resist motion and the drive from the electrical signal will be insufficient to overcome those forces. Then you start to get drop out of signal.
It is not just belief, it is empirical experimentation. It can easily be heard with the decay of sounds in a real acoustic environment and then the reproduction of that environment will often entail a truncated decay. The further that decay is captured the better the low level performance of the system and the better a system can capture that decay when there are other louder sounds also being played (like decay of a piano while the next notes are being played...with a real piano you can hear the decay of the previous notes even when the next notes are flowing onward...with hifi this is often difficult to hear).
"If one plays a recording of an orchestra where the live peaks were 105 dB, and sets the playback volume to, for example, peaks of 80 dB (a reduction of 25 dB), minute details will obviously be lost. This is common, and, to be expected. "
For sure, and why some kind of compensation to return things to a more lifelike level without actually turning up the volume is perhaps interesting. THis is because it is clear that wide dynamic range music loses interest when not played at the realisitic volume usually. However, there is a HUGE difference between systems on how well that 25db reduction is handeled and some lose much less than others. The Acoustats I had (Spectra 2200s) EXCELLED at retaining the detail of the low level information even when the peak level was reduced by a large amount and thus kept the music more interesting. The only speakers I have had that are as good are the horns i have now, Stax ESL F81s and AudioStatics. My Apogees, as good as they were, could not do as well in this test nor could the Infinity Betas or the Genesis VIs.
"You'll have to provide some evidence that the signal simply isn't there, and that it's not because the listener couldn't hear it."
I never said the signal isn't there, I said it is not being reproduced or reproduced increasingly inaccurately. It is also possible to get burried in the electronics noise floor as well, especially if the piece of gear has a high negative feedback that creates a signal correlated "noise" floor. This was demonstrated by Crowhurst back in the 1950s. Of course it is also possible that the listener is not able to hear it below a certain level but I think it is more often system related because in a good acoustic environment with live, unamplified music you can hear these things (or at least I can) clearly.
"I never said the signal isn't there, ..."
Yes, you did. You wrote:
"...subtle sounds will simply drop out because the system cannot generate such small responses. "
You're mincing words. "Drop out" and "isn't there" both mean it isn't there, and isn't that what you're saying?
First, you were talking about "the system", which you later modified to "the speakers". There's a difference.
This makes me wonder about all those reports and reviews of systems (er, loudspeakers, er, amplifieers, er, preamps, er, sources, er, interconnects, er...) where the writer talks about it reproducing the most subtle nuances and details of the recording, and having such wonderful "air". You know exactly what I'm talking about. They must have been playing the system at "concert level" (or louder?).
I've already asked you for data. I'll ask again.
If your assertion is correct and you have empirical evidence, cite a reference or two (preferably three), or point me to where I can find it. Anecdotal "evidence" doesn't count.
In any case, there are at least two factors which make the "problem" moot:
1. At "low" (define that) listening levels, subtle detail is lost in the room's ambient noise.
2. You still haven't provided evidence.
Let's not forget that audio level compression and limiting has been around since darn near forever. This is not news. The problem of low level playback compensation was addressed many decades ago, by many companies. Probably not to your liking or approval. Google "Gain Brain" and "Kepex" and "LA-4".
:)
By "system" I meant loudspeakers in that particular case but it can also apply for electronics but in their case it is more that the signal is dropping below either a true noise floor or a signal modulated one.
"drop out" and "isn't there" don't really mean the same thing. Drop out means something that is there but below the capability of the system to produce the sound. Isn't there means its somehow missing from the music content and that is not what I am saying.
"First, you were talking about "the system", which you later modified to "the speakers". There's a difference"
I agree, the confusion is first I was mainly referring to speakers but it can also apply to electronics although the reasons why are completely different between the two. I ended up sort of mixing the two together (in my mind they both impact the low level resolution and dynamics though).
"They must have been playing the system at "concert level" (or louder?)."
Often the case and many of my audiophile friends listen a lot louder than me.
"I've already asked you for data. I'll ask again.
If your assertion is correct and you have empirical evidence, cite a reference or two (preferably three), or point me to where I can find it. Anecdotal "evidence" doesn't count."
I'm sorry, what exactly am I supposed to be proving to you? Not that I am in any way obligated to do so since you have a keyboard and google as well.
Those compressors and booster are to compensate (inadequately because they are not level dependent) for the drop out that I speak. It is not just the hearing that is changing but the speakers and amp capabilities as well.
There is a good reason why a horn or electrostatic speakers continue to sound lively, detailed and with good ambient retrieval at far lower average listening levels than a conventional, low sensitivity speaker. This alone gives a clear indication that some speakers "drop out" low level sounds earlier than others. You can further easily hear the influence of the electronics on this with a given speaker. I am giving you firsthand experiment observations but I am surely not alone in this. Electrostats give a great insight into the effect of electronics on the whole low level resolution issue. Horns as well but some of them have colorations and resonances that mask information too.
"I'm sorry, what exactly am I supposed to be proving"
You are kindly asked to provide data/references which support your assertions.
Too bad that live sound is not typically EQ'd or "corrected" according to individual listeners acoustical vantage point.Because everyone hears live sound differently, it is therefore crucial that live music be loaded into our memories for proper reconstruction at some later point in time. For it is written, "The best hifi system is the one in my head."
And so it may be that the best hifi system transmits as much of the electrical signal as cleanly as possible while also helping us to construct or recreate some musically IDEAL vantage point, for ourselves. We must all find our own EQ curve. EQ according to memory.
Edits: 06/21/15 06/21/15 06/21/15
Clearly, there is some confusion here about the point of your question.
At the outset, your question seemed to be obvious and clear. However, several people seem to think that you're asking about loudspeaker frequency response curves. For our collective recollection, here is what you wrote:
*****
"Many of us use some sort of real-time sound level meter to position speakers and/or configure equalizers to get smoothest (i.e. flattest) frequency response at the listening position.
Yet a measured flat response at the upper end of the spectrum might sound subjectively way too bright and a measured flat response at the low end might sound subjectively too anemic. And I'm not talking about problems of calibration with the meter itself.
I remember an "ideal" measured response curve published many years ago by Bruel & Kjaer (or somebody)that addressed this. "
****
Re-reading your original post, it's clear that you're asking about the room curve, as averaged over some period of time, not the speaker's FR curve as measured in anechoic/semi-anechoic/gated conditions. Please correct me if I misunderstood your question.
And, so, yes, a roll-off at upper frequencies is desirable, and is one of the things which acousticians consider when designing or renovating rooms for music.
:)
In your post you say "And, so, yes, a roll-off at upper frequencies is desirable, and is one of the things which acousticians consider when designing or renovating rooms for music."
However, what if the room is acoustically dead or one's high frequiency hearing is beginning to roll off? Its much easier to correct in room response with a speaker whose response is flat across the entire spectrum than one that is rolling off in the high end.
In most rooms, the frequency response you hear at the listening position is dominated by the overall power response radiated by the speaker into the room and how it interacts with the room. The overall power response is never flat.
Suppose you design a speaker that is perfectly flat when measured at 1m distance on the tweeter axis in an anechoic chamber, which is a design target for some companies. Unless the speaker has perfect constant directivity (none does), the response off the tweeter axis will not be flat. The dispersion characteristics of most speakers lead to an off-axis response that slopes downward with frequency, with the slope increasing as you move further off axis. Hopefully it's a smooth curve without any major peaks or suckouts, but few speakers achieve that. And the peaks and suckouts vary depending on whether you are moving off axis horizontally or vertically.
Also, as you move out from 1m distance, the on-axis response will change too. It will become generally down sloping, with some peaks and dips. This change with distance occurs even under anechoic conditions.
Good speaker designers pay as much attention to the off-axis response as the on-axis response, making trade-offs between flatness of the on-axis response against the shape of the total power response. If you chase perfectly flat on-axis pseudo-anechoic response and take what you get off-axis and at distance, the result ends up sounding colored in-room.
Now suppose you were to design a speaker that is both flat on-axis and you try to control off-axis response by limiting directivity as much as practical. The result will have a flatter power response, but will also sound unnaturally bright. Because music isn't being produced with that kind of target response in mind.
I concur that there is no such thing as a perfectly flat speaker. I just don't like designs where they purposely try and roll off the high end.
As has been pointed out several times in this thread, flat does not sound correct. Everybody designs for a roll-off. The question is how much top end roll-off is right in a speaker, and there is no objective answer because the end result is dependent on external variables such room, listening distance, positioning, and so on.
That's where the B&K and similar curves come into play. The B&K curve is one attempt to capture what response sounds the most natural. A designer who accepts the B&K curve as a reference will try to achieve a power response that follows the curve in a typical room. Of course, deciding what is "typical" is guesswork.
You seem to be hung up on thinking that there is an objective standard that you can measure loudspeaker frequency response against and there really isn't.
One might be forgiven for thinking that an objectively good speaker is one that measures flat on-axis at 1m in an anechoic chamber, but that is neither necessary or sufficient. It is a somewhat arbitrary design target. I can design two speakers that both measure flat on-axis at 1m pseudo-anechoic, but with very different power responses, one that rolls off slowly with increasing frequency which sounds too bright and one that rolls off rapidly and sounds too dark.
Similarly, I could design a speaker with an intentional rise or dip in the on-axis 1m anechoic response which compensates for a falling or flaring off-axis response such that the response at the listening position 3m away is smoother and flatter than if I had designed it with a flat on-axis 1m anechoic response.
"Good speaker designers pay as much attention to the off-axis response as the on-axis response,"
Seriously?!!
And this is news to whom?!
:)
"...or one's high frequiency hearing is beginning to roll off?"
What did you say? Could you speak up - I can't hear you. ;)
It's well-known that a reverberation characteristic which is somewhat longer at the low end and shorter at the high end generally provides a more desirable room sound for music. It gives a greater sense of "fullness" or "body" to the sound. Whether or not this is related to our reduced sensitivity to lower frequencies, I don't know - I haven't ever studied it from that angle - but I'd suspect that it is.
With regard to acoustically "dead" rooms, that's an interesting point. If you've ever been in a really nice studio control room, you know what a pleasure it can be, even just for conversing with another person - forget the music, just talking sounds great! Here'e the funny thing, in a small-ish, dead-ish room, there really isn't a "reverberation time" per se. This is because the "reverberation time" is so short that the sound never gets to what we would call a "steady-state" level from which we can measure an RT60. This is not to say that the quality of the reverberant field isn't important - it is! - but it takes another set of skills to do small-ish dead-ish rooms well, and that's something which I haven't done in YEARS.
Hearing loss: Oddly enough, adding more high frequency content to the room makes things worse, not better. What we want is more detail, and adding more high frequency energy to the reverberant field doesn't accomplish that. This brings up the issue of speaker output, and how it affects the sound we hear, in the room we're in.
Does that answer your questions?
Based upon your questions, you don't appear to be a neo-audiophile.
Fess up. :)
:)
I'm a crusty objective engineer..an audiophile's nemesis or nightmare. :)
My definition of speaker detail would be a speaker's ability to faithfully reproduce the signal without adding a voice of its own by either smearing phase relationships or by emphasizing or de-emphasizing (if that's a word) a single or a band of frequencies. Distortion also plays a significant role in this and I'm going to be looking at Floyd Tooles work on linear and non linear distortion and how humans perceive it. I want to understand this.
Adding more high frequency energy in a deadish room would help make the high end sound more available but in a lively room would hurt the high end. I would think that its easier to correct for a flat response whether you have to emphasize or de-emphasize the highs without affecting neighbouring frequency bands.
Its nice debating this with you BTW. :)
INMATE 51
Yes, that's exactly what I meant. Whatever the anechoic response of the speaker coming out of the factory, if the high frequencies measure flat at the listening position, you are likely to find the system subjectively too "bright." The B&K curve (thanks, DAVE K: that's the one I remembered)-- to my ears, at least -- is the better "shoot-for" curve in a normal listening room.
Great discussion, guys and gals. Thanks for all the responses.
Back in the olden days of audio, an ideal freq response could be calculated if you multiplied the low freq by the high, you'd get 400,000. For instance 20 X 20 k, 30 X 13 k... I am not sure have valid it is, but it was a general rule of thumb.
Read and be edge-ma-cated
https://en.wikipedia.org/wiki/Equal-loudness_contour
Since hearing is purley subjective, what's good for the goose many not be good for the gander so its best not generalize too much about this. That being said, I prefer a flat as possible response from the speaker (ignoring room effects) even if it makes some recordings sound bad. Why? I like to hear differences in the abilities of the recording/mastering engineers to see if they did their homework. I want the speakers to play the truth rather than smear the sound to make it more palatable. Thats what I look for. Your mileage may differ.
It is an observation of how we hear at lower levels - whether that is a live event or from our music systems.
Not lower levels only but across the spectrum. Human hearing is most sensitive in the midrange but is less sensitive on either side of midrange.
I looked at your F-M Wiki link, and thought it would useful to include a more detailed original version. This is from Knudson and Harris's "Acoustical Designing in Architecture", published by John Wiley & Sons, 1950. I highly recommend it.
Did you note the most important point about Figure 2.5?? This famous curve is only valid for PURE TONES. So, real music might look completely different than these famous curves but god only knows at this point what that might look like.
nt
That's the way humans have always perceived sound!
That doesn't answer the question about an ideal room curve.
Nothing more
There's no need to compensate with how humans hear. What varies in our innate perception has to do with level .
Perhaps you noticed the steeper curves as level falls.
Did I say there was a need to compensate how humans hear?
Perhaps I did notice...Your point?
in observing that variations in the F-L curves are due to level .
"It is an observation of how we hear at lower levels - whether that is a live event or from our music systems. "
My bad...lower levels, not lower frequencies.
.
... Yamaha receiver makes that possible?
As long as your not driving the unit into clipping, then the Yammmy delivers it in spades without coloration. I find it both ignorant and tiring that audiophools think that an AVR cannot reproduce the soundtrack faithfully.
Clearly you are not up to date on the latest research into the impact of electronics distortion on sound quality...say from the mid 20th century onwards!!
Read some of the work by CHeever, Geddes and Crowhurst (from the 1950s and 60s no less). Also note that the problem with amplifier distortion giving a "sound" of its own has been known for a long time and has been the subject of research on and off for a long time now. D.E.L. Shorter proposed that it was the square of the amplitude of the harmonic and even that was found not to be extreme enough such is the non-linearity of perception.
I am sure that the Yammy does NOT produce color free sound at any level. Most of it will be insidious but some things like lifelike dynamics will be obviously curtailed...its just an outgrowth of the design philosophy they use in making their SS amps.
A Yamaha receiver isn't expensive enough to produce high quality sound. You have to spend enough to get above the riff-raff, preferably with tubes (ANY tubes). THEN you're talkin' good sound, and for Pete's sake, DO NOT EVER use a tone control.
Hang on...
I think my tongue darn near came right through my cheek. Gotta go.
;)
too funny!
Everyone thinks I'm strange except my friends deep inside the earth
nt
So 3db, evidently I'm receiving your replies telepathically and thinking they're my own. Man, wish I coulda done that in high school. Well, not YOUR thoughts specifically, but you get it. Dave
Everyone thinks I'm strange except my friends deep inside the earth
LOL
Edits: 06/17/15
ROOM CURVE.
Fletcher who?
Tonality is important, but it isn't everything.
What usually gives the illusion of reality is nailing the transients and the pitch definition (being able to distinguish the complexity of the tone with relation to each other).
If you don't have good pitch definition, and transient reproduction nailed, tone will only take you so far.
============================
As audiophiles, we take what's obsolete, make it beautiful, and keep it forever.
Hey! I have a blog now: http://mancave-stereo.blogspot.com or "like" us at https://www.facebook.com/mancave.stereo
I saw this article on the "house curve" a few years ago and thought it might fit in here:
Nt
at lower levels both highs and lows sound lower than they measure. I suspect, assuming a well recorded piece, that the correct balance would then be playing back at the level at the ears the mastering engineer heard at his ears as he did his job.
You're right, but it's only a start.
To hear it as the mastering engineer intended, yes you would need to play it back at the same level. But you would also need a system with the same frequency response. And you would have to consider the mixing engineer too, because the playback level it was mixed at and the response of the mixing system surely influenced the mix. And surely the monitoring level and response of the monitoring system influenced the recording engineer's decisions too. Given there is no standard listening level or frequency response in the recording industry, and recordings are frequently put together by chain of multiple engineers working at different levels in different systems, it's no wonder why natural sounding recordings are hard to come by. It also helps explain why better recordings often come from small labels with a team of a few individuals doing everything.
One reason why there are mastering engineers is to balance/tweak the sound which the mixing engineer created, and to make the sound quality and loudness of every track "match" with the others, and to sound good/great against competitors in the marketplace.Another reason is cost. Nobody in their right mind would spend the money to do that job while sitting in a recording studio control room full of stuff, with an expensive studio on the other side of the glass. Mastering rooms are "bare bones" and specifically designed for a different specific purpose.
Lastly, there is no "one proper" level or EQ setting to hear a recording "as the mastering engineer intended". Another part of their job is to make the recording sound good in a variety of playback environments at a variety of levels.
:)
Edits: 06/17/15
I wasn't trying to debate whether an independent mastering engineer adds value to the process.
I was only trying to emphasize that there are no standards to help ensure good results when the product is being developed by multiple engineers working in different places on different systems.
If frequency response and monitor level were standardized in the music industry like they are in the film industry, then in theory the mixing engineer and mastering engineer's decisions are primarily artistic. Whereas today they can be influenced by perceived differences in loudness and frequency balance that arise due to working at different levels on systems with different frequency response.
Bob Katz and Floyd Toole have articles available online that discuss the problem and solutions in more depth.
Something of a rarity, but I know that recording engineers or other pros involved in the production of a recording sometimes do specify a *proper* volume level for listening.
Read the liner notes in some of John Marks' CD recordings sometime. He definitely does believe in a proper level for the recording at hand and he definitely specifies what that level is.
d
Edits: 06/16/15 06/16/15 06/16/15 07/06/15
+9 dB boost over the bottom two octaves and a flat shelf from 100 to 2k? Now that would sound like ass.
But based on the Sean Olive papers, I don't think that's actually the JBL Synthesis target curve.
and sacrifice any notion of a realistic image at the same time.
d
Edits: 06/16/15 06/16/15 06/16/15 07/06/15
If you want real imaging, you need multi-channel speakers
What exactly is a "multi-channel speaker"?
Hopefully, you first name isn't Ed. :)
Nt
"Yet a measured flat response at the upper end of the spectrum might sound subjectively way too bright and a measured flat response at the low end might sound subjectively too anemic. And I'm not talking about problems of calibration with the meter itself."
If this is the case, I believe there is some sort of distortion(s) in the signal (chain) causing the perceived brightness, not the FR itself.....
"I'd love to resurrect this topic, especially since digital correction is such a hot item right now."
Digital correction IMO causes more problems than it solves.......
A friend who has designed speakers commercially told me that flat response sounds bright. The solution is a 'flat' curve that slopes down ever so slightly with increasing frequency.
If I recall correctly the reason for the slight response decline was that the power response of a flat speaker made the speaker sound bright.
I think part of it has to do with trying to get true full-range sound with speakers whose boxes and/or drivers are undersized. The excessive excursions of smaller bass drivers often result in both harmonic and Doppler distortion, which does add amusical components to the upper end of the audible range.....
With larger boxes and drivers, the bass is linear, allowing the harmonics of all instruments to be heard in a more natural state. I've found that lack of bass linearity causes the "brightness" that is too often blamed on extended HF or flat FR.
I've even noticed the impact of bass linearity on the top octave.... IMO, high-frequency clarity is almost a lost art in sound reproduction. Hearing the sweet "sparkle" of the triangle in the midst of full orchestral forces often brings goosebumps inside a concert hall, but is one of the most-difficult things to capture and get right in home audio.
The reason why a rolled off response sounds best is that recordings are made to sound good in the mastering studio on the mastering studio's speakers. Most of these have high frequency roll-off. Therefore a recording played back on an audiophile system with flat response will sound brighter than the way the mastering engineer heard it, that is to say it will be too bright (assuming the mastering engineer did a good job making the recording).
The question then arises: why does the mastering engineer have a speaker that is rolled off? The answer is that the mastering engineer wants his recordings to sound good on a variety of playback systems. These range all over the place, so it is a bit of a crap shoot. However, if his recordings fall near the center of the rolled-off vs. too-bright distribution of recordings then he will have the best chance of his recordings sounding good on many systems. The historical fact is that most recordings sound best when played back on systems that have slightly rolled off high frequencies. Therefore it is helpful to the mastering engineer if his playback system is slightly rolled off. When setting up his mastering studio, a first rate mastering engineer will dial in the high frequency response of his monitors by listening to a wide range of recordings and picking a reference set used to set up his monitors.
Tony Lauck
"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar
"The question then arises: why does the mastering engineer have a speaker that is rolled off?"My answer is that the digitization of audio has essentially forced such practice.... Prior to the digital age, extended HF (beyond the audible range) was often highly desired, some even claiming it enhances how we perceive the sound within the audible range.... (Even Radio Shack was marketing ribbon tweeters and other extended HF products in those days.)
This totally changed when the bandwidth limited CD displaced vinyl as the consumer medium of choice. (And with MP3 dominating current music sales, the need for extended HF seems to have gone away completely.)
With the comeback of vinyl, if future digital electronics (computers, entertainment products, etc.) can tame the RFI emissions which I think brought down quality sound reproduction at all levels, we might have a resurgence of audio products with extended HF. Not to mention maybe even studio products that do likewise.
"The answer is that the mastering engineer wants his recordings to sound good on a variety of playback systems."
I don't think using rolled-off speakers would be an ideal conduit to make recordings sound good on a variety of playback systems. Even playback systems of the Best Buy variety.... If one doesn't know what's up there, how would he know if it will sound OK on other systems?
Maybe this is why we've had recordings with strange HF artifacts in recent time. If the drivers are excessively rolled off, and the mastering engineer doesn't hear the top octave, he might not be aware of HF artifacts that could make it to the final released product.
The mastering engineer should be able to hear everything, good, bad, and ugly. He would then be able to better determine what would be best for a variety of systems.
Edits: 06/17/15
In my original post, I provided a link to Bob Katz's book. It has a whole chapter on how to set up a mastering studio which goes into the question of high frequency roll off in the monitoring.
This has nothing to do with the high frequency response of digital vs. tape recorders or vinyl. This has to do with rooms, microphones, speakers, acoustics, and psycho-acoustics. All the same issues of frequency response, too bright recordings, etc. existed back in the early 1960's when I was listening to pre-recorded stereo tapes and mono LPs.
Tony Lauck
"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar
"The reason why a rolled off response sounds best is that recordings are made to sound good in the mastering studio on the mastering studio's speakers. Most of these have high frequency roll-off. Therefore a recording played back on an audiophile system with flat response will sound brighter than the way the mastering engineer heard it, that is to say it will be too bright (assuming the mastering engineer did a good job making the recording)."
Tony, you seem to be a bright guy, but you, along with others, are on the wrong track here.
The question is not about speaker FR curve, but rather about ROOMS and how we perceive the sound in them. A rolled off upper frequency curve is preferable - FOR THE ROOM SOUND - which is what the OP asked about.
:)
I was talking about frequency response of speakers in rooms at the listening position . Speakers are designed to be used in rooms and their frequency response outside of a room is irrelevant to an audiophile or recording engineer (although not to a speaker designer). The in room response of a speaker will vary according to its position in the room as well as the listener's position in the room.
This does not describe why frequency response in the listener's room at the listening position needs to be rolled off. This depends on understanding how recordings are produced.
Tony Lauck
"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar
| This does not describe why frequency response in the listener's room at the listening position needs to be rolled off. This depends on understanding how recordings are produced.
Specifically, why do you think the frequency response needs to be rolled off? What's important mechanism, and what isn't?
How much validated by experiments?
This isn't a rhetorical question. I think we all now agree on the phenomeon but I want to know other than "it sounds better", why? What part of measurement and playback which, at least superficially seems to say that we have reproduced the same thing, doesn't actually reproduce the psychoacoustic same perception?
Presumably if we cloned everything about the original and transplanted a new human into that recreation, like an acoustic holodeck, then some version of "flat reproduction" should be correct.
My record library consists almost exclusively of recordings of acoustic music, a mixture of studio recordings and recordings made in concert halls with or without an audience. If I play these recordings on a system set up for flat high frequency response (as measured at my listening position) these recordings are heavily skewed to the bright side. Some highly regarded recordings, such as most of the Mercury Living Presence recordings, are so bright that they are painful to listen to, the equivalent of "fingernails on the blackboard". None of the recordings in my library sound dull with my system set up flat. When I adjust the high frequency response of this system so that the recordings range from slightly dull to slightly bright, nearly all of the recordings in my library sound very good, and the ones in the middle sound excellent. None are painfully bright and none are boringly dull.
I know what live music sounds like from decades of going to live concerts. I know which recordings in my library are the best from reading reviews and years of experience listening to many of these on different systems. I know that, as originally set up with "flat" cross-over settings, my system was horribly bright. I measured it's response at my listening position and it was flat from 1 kHz up to 20 kHz, which was as high as my calibrated microphone went.
After adjusting the high frequency response (by turning the high frequency control for the tweeters on my left and right speakers half-way down) I got a setting that sounded approximately correct, consistent with live music and consistent with how these recordings sounded on other systems. I measured this and found that there was a gradual roll-off starting around 1 kHz and ending up down -3.5 dB at 10 kHz, leveling out at a maximum attenuation of about -4 dB at 12 kHz and up.
I already explained why recordings tend to be too bright on flat systems. That's because that's the way they are made. If you want to know why they are made that way, you would have to talk to the people who made these recordings (most of the great engineers are no longer with us). While you are at it, you can also talk to the successful speaker designers, who won't sell many recordings if their products were flat because customers don't buy products that give them "bleeding ears".
Anyone who wishes to confirm this for himself will need to meet a few requirements:
1. They must have a system that is full range for the music being played.
2. They must have the ability to accurately measure their system response at the listening position using calibrated instruments.
3. They must have a record library with a wide range of acoustic music
4. They must be intimately familiar with how live music sounds by attending numerous concerts of different types of ensembles in different halls from a variety of listening positions in the venues.
5. They must have the ability to compare their memory of live events with what they hear when they play recordings of similar ensembles on their system.
6. They must have some way to adjust the frequency response of their system, at least as flexible as traditional tone controls.
7. They must have the ability to listen critically and identify small differences in frequency response while listening to music, down to at least the +- 0.5 dB range over 1/3 octave bands.
None of these requirements should be difficult for a competent audiophile.
Tony Lauck
"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar
"customers don't buy products that give them "bleeding ears"
Very thoughtful response, Tony. Nice contribution.
But I'm not sure manufacturers don't mind adding a little sizzle to their steak. It's called "detail." I hit the CES and Rocky Mountain audio shows every year, and too many times I don't make it more than three feet into the demo room before I rush right out again. OW! OW! OW!
Perhaps its because there is no blood left in the ears of the younger generations as a result of attending a few rock concerts or from having headphones permanently welded to their skulls. Or maybe they figure that the real buyers of their stuff are old guys (like me? us?) who need a bit of boost in the high freqs or we can't hear them at all. But that's all just speculation.
I haven't destroyed my hearing by abusing it. When I was 19 I could hear 21 kHz. Today, at 71, I can still hear 13000 kHz. Live music doesn't sound dull, and I can hear tonal differences in music when cutting or boosting frequencies above 15 kHz, even though I can't hear sine waves at these frequencies any more. The mind provides needed recalibration of frequency response providing that one's hearing changes gradually. If there is sudden damage due to accident or illness then this may not happen.
I have no use for equipment that is marketed to people who listen to non-acoustic music. The existence of these people and the existence of this music is one of the reasons why there has been little real progress in audio over the years. If I were appointed "Musical Dictator of the Univers" my first edit would be the banning of all musical instruments that are not operated on human power. (This would even include banning pipe organs, unless they were hand pumped.) :-)
Tony Lauck
"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar
After your ban on non-acoustical instruments becomes effective, what will we do with all of those people who once liked them?
Oh man, my head hurts. Put yourself (speaking to no one in particular, mostly all you cats that are so well read in this stuff) in my shoes, just a regular joe who enjoys hi-fi/stereo/music & equipment. Now I'm trying to grasp a little of what this thread is saying & how it applies to me or my listening & understanding music. After 3 or 4 posts, I felt as lost as I would be listening to Einstein explaining the relativity deal. My head has shut down. So maybe I'll just go enjoy some music & let you guys debate whatever it was this started with. ** BTW, I'm not being flip, I honestly admire all you who understand this stuff. I just accept that I ain't gettin' it. Carry on men, Dave
Everyone thinks I'm strange except my friends deep inside the earth
| If this is the case, I believe there is some sort of distortion(s) in the signal (chain) causing the perceived brightness, not the FR itself.....
I think the evidence (starting from B&K and the recently posted work from Harman, and what Anthem does which it says is from NRC research) is against this. It's the frequency response: Down With Flat!
Your ear & brain compensates and doesn't think that in-room flat response sounds right. And experiments show anechoic flat isn't flat in the room.
To reverse Duke Ellington: If it sounds bad, it *is* bad.
J. Gordon Holt, 1985:
Many times in past years I have been impressed by the incredible flatness of the measured high-end response of some speakers: almost like the proverbial straight edge out to 15kHz, and sometimes beyond. In every such case, I have been equally amazed at how positively awful those loudspeakers sounded—so tipped-up at the high end that could not enjoy listening to them. (They aroused a deep nostalgia for the days when preamps all had tone controls.)
Nor am I the first to have observed that an objectively flat high end sounds tipped up. Ever since acoustical engineers started using equalizers to "voice" recording studios and monitor systems, they have observed this marked disparity between what measures flat at the top and what sounds flat. They were all ultimately reduced to pulling down the whole high end and—Heaven forbid!— adjusting it by ear. No one seems to know why this is so, but the important thing right now is that it is.
Since the human ear's sensitivity changes with loudness, adjusting the system at one level will change the moment the music gets louder or softer.. IF you are relying on meters and response curves.
I certainly would think someone should have, by now, made an 'enlightened' loudness contour switch..
One which mimics/adjusts to the natural response of the human ear.
I have always found a subwoofer set to be perfect for chamber music is way way too loud the moment Rock and Roll is put on. Or, if adjusted for Rock and Roll, then the Classical music and Jazz response of the sub sucks.
It is all about perceived loudness
I find that the lowered sensitivity of the bottom octaves occurs whether I'm listening live or at home. In other words, *correcting* the bass level when music is played back at lower levels (or what my brain thinks is a greater distance from players) sounds unnatural to me.
Live (unamplified) performers don't have loudness buttons nor do I find them useful for my music systems.
You hit the nail on the head. Live music must follow the same loudness rules, so classical basslines playing at piano level are supposed to be really quiet. So assuming the piece is properly recorded, why would you want to boost the level just to satisfy some wrong headed notion of "correctness"?
The way things are going, 20 years from now, audiophiles will be debating which tone control and loudness contours are ideal.........
This is why I cringe at the "digital room correction" thing.... There was once a time where doing this in the analog domain was frowned upon, even though it was a lot more transparent sonically than digital room correction.
with the HT system (where bass traps are not used), I use a mild amount of attenuation using the processor's parametric EQ to flatten out two peaks centered between 90-100 hz.
Annoying boom gone from the room.
"I find that the lowered sensitivity of the bottom octaves occurs whether I'm listening live or at home. In other words, *correcting* the bass level when music is played back at lower levels "
Fine, great. But you are also on a different topic than what the OP is talking about.
:)
the topic found in the post from Smelly Socks to which I replied.
Had I replied to the OP, I would agree with your observation.
To recap:
Smelly Socks is wrong about the topic of his response
I find that Smelly Socks is also wrong about his wrong response.
Does that clear things up? :)
Smelly_socks is exactly right.
"Equalization" is meaningless, since it's volume dependent and recording dependent.
This is why they used to have tone controls.
My goal is a realistic reproduction of the experience of listening to live acoustic music in a concert hall, recital hall or jazz venue. The recording will provide an acoustic perspective as to the size of the hall and the listener's position. Were the listener to listen to an actual (or imagined) concert from that position, he would hear a range of volume levels (from ppp to FFF) and tonalities as produced by the musicians playing their instruments. Realistic reproduction requires that the volume and tonality the listener perceives from his listening position match what he would have heard in the (real or imagined) live performance, whether the musicians are playing loud or soft.
The psycho-acoustic effects on tonal balance of volume do not come into play if the volume levels are correctly matched, because the volume related hearing sensitivity would be the same for both the live and the reproduced music. If the combination of record equalization and playback equalization match and the listener adjusts his volume control correctly then he will get a level of realism throughout the entire recording.
There are two provisos. First the recording must be undistorted and uncompressed, so that a constant volume control setting will apply throughout the entire recording. Second, the playback chain must be capable of undistorted output of the full range of musical sounds produced by the instruments at the required volume levels. If these are not met, then realism is hopeless and the recording and/or playback system are not suitable for anything but background music. In this regard, a system that is adequate for realistically reproducing solo harpsichord music may not be able to reproduce organ music recorded in a large cathedral or a Mahler symphony in a large concert hall.
I have found that playback systems that use stepped volume controls with 2 dB (or larger) steps do not permit sufficiently accurate matching of levels for many recordings. Volume differences of about 0.5 dB are audible, so a stepped volume control with 1 dB steps is (barely) adequate in this regard. Frequency response differences in the amount of 0.5 dB (or less) are readily audible over various frequency bands. The traditional bass and treble tone controls lack sufficient flexibility to correct errors in recording equalization or speaker and room response, among other reasons because two knobs are not enough.
Tony Lauck
"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar
You'd think that, after 50 years of Audio, Stereo Review, Stereophile, TAS and other ragazines, we wouldn't still be having this conversation.Long ago, I recognized that Rule #1 about these mags is that they never get past "Audio 102" and into more advanced info. That's one big reason why I haven't bought one in a loooong time - there's nothing new. But I digress.
The OP was asking about a room curve. This doesn't vary with level. You and Socks are talking about an entirely different topic.
Some of you guys really need to pick up a book on acoustics and read it, as I said in this post:
http://www.audioasylum.com/audio/general/messages/70/700456.html
... and which was the topic of my thread "Books Which Every Audiophile Should Own" (or something like that).
:)
Edits: 06/16/15
The B&K curve can be found here:
http://www.bksv.com/doc/17-197.pdf
Harman did a study and came up with a similar, but more tilted curve:
The Subjective and Objective Evaluation of Room Correction Products
There is no right answer. As Toole laments in his book Sound Reproduction, there is no standard governing the frequency response of systems used in music production. The film industry has its X-curve but the music industry doesn't have an equivalent.
The B&K curve is a pretty good place to start however.
I've been demoing an Anthem MRX 710 receiver. It estimates a "room gain" which effectively boosts the target in the bass for in-room response.
There was an interview by the designer who said that it's your ear & brain that automatically "normalizes-out" (my interpretation) the effect of the room. (like the way your brain normalizes color balance that photographs don't).
So if you equalize something to sound flat anechoically or outside, then it will be tilted towards bass indoors, and this is how it should be.
If it's equalized to be flat indoors, it will sound lousy.
Yes, flat anechoic response is also a good place to start. However, it isn't necessarily correct either. The room contribution depends on the radiation pattern of the speaker, so you can equalize two different speakers to be flat on-axis under anechoic conditions but they will have different in-room power response at the listening position. Speakers with greater directivity will sound brighter. When experimenting with various flavors of DSP equalization, I've always found there's a tradeoff between optimizing pseudo-anechoic response vs. optimizing overall power response.
Many people who use RTAs (real time analyzers) shouldn't be using them, because they're clueless about what they're measuring and whether or not the measurement means anything useful.
Having gotten that off my chest...
There are desirable and undesirable "curves". Which "one" is right for your situation depends upon several parameters, including:
1. Size of the room.
2. Type of sound being produced.
3. Type of sound being reproduced.
4. Intended audience.
This has been discussed at length in the literature for many decades.
Open a book. Leo Beranek's "Music, Acoustics & Architecture " might be a good place to start, but may be over your head. Still, read it... it'll be a great source for you - kind of like jumping into the deep end.
Aside: Sadly, I'm no longer amazed at how little knowledge "audiophiles" have about their interest, and how little interest they have in furthering their knowledge.
:)
The frequency response you hear is a combination of the frequency response of the recording and the frequency response of the playback. In addition, for various physical and psycho-acoustic reasons, there is no consistent way of measuring frequency response from the perspective of how it is perceived by people (musicians, recording engineers, producers and audiophiles). One of the problems is that the acoustics of recording venues and llistening rooms vary all over the map. If a system is set up to a listener's taste while playing one recording it may not sound so great on a different recording. If the system setup is changed to make the second recording sound better, the first recording may sound worse. (Example: a bright system may make a dull recording sound good, but it will make a bright recording sound harsh, even unlistenable.)
There are no sound standards for how recordings are made. Unless you want to be continually changing or tweaking your system you will have to compromise. Find two dozen recordings that you know are good. These should include a variety of musical genres that interest you and a variety of recording styles and record labels. Tweak your system so that these all sound good or at least represent a suitable compromise. Some outliers may be a bit dull, while other outliers may be a bit bright, but all should be musically enjoyable if you don't have any poor recordings in your sample and if you've done a good job selecting and setting up your system.
I achieved a house curve that is flat from below 30 Hz up to about 1 kHz and then rolls off gradually from 1 kHz to about 10 kHz, after which it stops rolling off. Response is down about 3.5 dB at 10 kHz at my listening position. This works well with acoustic music, including classical music and jazz. Of course it is still necessary to make volume control adjustments for each recording and sometimes I move closer or further back from my speakers to adjust the sound stage. I listen at natural concert volumes. Were I to listen at lower volume then perhaps I would have preferred some bass boost.
Tony Lauck
"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar
You are remembering the B&K curve. Start at about 200-300Hz and decrease the amplitude about 1 dB per octave. So it will be down about 5-6dB when you get out to 16-20 kHz. There are other similar "house curves"
Go listen. Then select your 5 or 10 most absolute favorite loudspeakers - look at whatever specs, graphs etc you like and then find the correlation.
Then select 5 or ten of your least favorite (oh heck let's just call it absolute horse dropping sound) and look at whatever specs, graphs etc you like and find the correlation as to what is going to sound like dredge.
But people respond to different things - some people like bright loudspeakers - some people don't seem to need dynamics others like some added warmth - some want transients more than decay - others want tone over soundstage. I have never seen a frequency plot that told me whether I would like or dislike the result. And people with decent hearing should be able to tell if something is "good" or not by listening.
I have never seen a frequecny response that let me know I would like the speaker, but I have seen some that told me I wouldn't like them.
Beatnik's stuff http://web.me.com/jnr1/Site/Beatniks_Pictures.html
Like or not like are rather vague, as there are degrees.
I certainly agree I have seen speaker measurements which tell me I would not like a speaker, or not like it well enough to consider.
I have other speaker measurements which indicate that in a careful set up, I probably could get sound I would like. I had a pair of the old Kef 104 speakers, and they could sound very fine in a suitable set up, but needed very careful placement. The listening window response was very even, with mild downward trend, but farther off axis, the response was not so even.
On the other hand, speakers with an even response even quite far off axis are likely to sound very good. I still prefer a slight downward trend above the midrange, such as with my current speakers.
I cannot possibly listen to every speaker, and I use measurements as a screening tool to eliminate unlikely candidates and point me toward likely ones. And then, I try to find them. When I audition them, I use a variety of recordings, some new, some old, which seem to indicate significant differences between speakers. For consistency, I include a few which I have used for decades. And of course, I want to try out the final candidate(s) at home.
This has worked out quite well for me. The only problems I run into here are 1) I screen some speakers favored by some here, and 2) some insecure people are upset that I find speaker measurements to be of any use.
-----
"A fool and his money are soon parted." --- Thomas Tusser
Post a Followup:
FAQ |
Post a Message! |
Forgot Password? |
|
||||||||||||||
|
This post is made possible by the generous support of people like you and our sponsors: