Concerning compensation of HRTF i have a simple
question:
Assuming the mastering of a certain recording has been
done using a stereo setup, would we also assume the
mastering engineer would have already accounted for
HRTF during mastering ? Just by aiming for a balanced
recording and "well defined" phantom sources due to his
auditve impression ?
If yes, an additional compensation in the homely speakers
used for reproduction would be an "overcompensation".
Please correct me if i am wrong.
Given the mastering engineer uses the latest "Orion"
with integrated HRTF compensation by a shelving filter,
the recording will sound too bright if reproduced
on a flat speaker at home.
It would take the same "Orion" version to reproduce
the recording in a balanced manner ...
I cannot help myself, but here is something going wrong
IMO. Maybe Mr. Linkwitz uses his own recordings as a
reference ...
That a slightly falling frequency response is preferred by
most listeners seems to be backed from different studies
too, but i am not sure whether this relates to HRTF
compensation when reproducing commercially available
recordings.
Kind Regards
Last edited:
These?
An externally hosted image should be here but it was not working when we last tested it.
(from here)
dave
Concerning the "non flat response" topic, I still suspect SL - as much as I admire him and his work - is on the wrong track here. The explanation for his observations put forward by JohnK, Rudolf and others (that it´s really about off-axis behavior in the HF area) still sounds much more plausible to me than SL´s take on it. The point LineArray is making is - in my book, at least - another convincing argument in the same vein.
The falling response curve people seem to prefer is for all we know a function of absorption and diffusion increasing with frequency in typical listening rooms, so decreasing HF output beyond flat should result in a non-optimal in-room response - if, and again this is in my view the main point of the whole discussion, the off-axis response is well behaved.
Just my 2 cents.
The falling response curve people seem to prefer is for all we know a function of absorption and diffusion increasing with frequency in typical listening rooms, so decreasing HF output beyond flat should result in a non-optimal in-room response - if, and again this is in my view the main point of the whole discussion, the off-axis response is well behaved.
Just my 2 cents.
Read Zaph Audio site
There is always tension between those who want flat and those who want something different. I am torn being a INFP on the Meyer's Briggs and yet knowing that a lot of subjective talk is nonsense (ie. Stereophile). I try to get my speakers as close to flat as possible by design and then tweak the settings on my pre-pro to my liking. As John Kruttke (sp? I think that's his name) has posted on his web sit "Zaph Audio," sound quality can be measured in every aspect and talk about subjective response is mostly due to a lack of understanding of the technology and what the sound indicates as far as measurement (he says its due to laziness, but I think that is too harsh). He is obviously not an INFP but I think what he says results from understanding acoustics. I really do find the whole language of subjectivism revolting what with terms like "air" or "quickness" and so on. Speakers are reproducers, they are not the source so what they are able to reproduce should be as flat as possible starting out in order to deliver all that is on the recording. Perhaps the mix has too much highs or too much bass or perhaps the delay or phase is wrong or perhaps there is a lot of second order distortion giving it that tube sound and therefore you need to adjust those parameters you can in your pre-pro. I personally prefer the sound of tubes, or did when I had a Stereo 70 years ago, yet I know it was not accurate.
There is always tension between those who want flat and those who want something different. I am torn being a INFP on the Meyer's Briggs and yet knowing that a lot of subjective talk is nonsense (ie. Stereophile). I try to get my speakers as close to flat as possible by design and then tweak the settings on my pre-pro to my liking. As John Kruttke (sp? I think that's his name) has posted on his web sit "Zaph Audio," sound quality can be measured in every aspect and talk about subjective response is mostly due to a lack of understanding of the technology and what the sound indicates as far as measurement (he says its due to laziness, but I think that is too harsh). He is obviously not an INFP but I think what he says results from understanding acoustics. I really do find the whole language of subjectivism revolting what with terms like "air" or "quickness" and so on. Speakers are reproducers, they are not the source so what they are able to reproduce should be as flat as possible starting out in order to deliver all that is on the recording. Perhaps the mix has too much highs or too much bass or perhaps the delay or phase is wrong or perhaps there is a lot of second order distortion giving it that tube sound and therefore you need to adjust those parameters you can in your pre-pro. I personally prefer the sound of tubes, or did when I had a Stereo 70 years ago, yet I know it was not accurate.
Last edited:
Let's get something straight: you can't reproduce in your music room in principle (and certainly not in practice) the aural experience of sitting in Carnegie Hall (or anywhere else including your own music room*) with two loudspeakers. You can't reproduce the sound of the studio or the mixing room.
I suspect that means there is no such thing as "flat" if you can't make something sound like something else which is flat.
I wish people would stop fantasizing about engineering "solutions" to psycho-acoustical problems.
Now let's get on with creating systems that we can enjoy by whatever different criteria we have, including "sort of feels to me like Carnegie Hall...." You have to figure out what variables matter in achieving those criteria.
No useful and universal way to define "flat."
*People with long memories will know what kind of demo-proof I am echoing here.
I suspect that means there is no such thing as "flat" if you can't make something sound like something else which is flat.
I wish people would stop fantasizing about engineering "solutions" to psycho-acoustical problems.
Now let's get on with creating systems that we can enjoy by whatever different criteria we have, including "sort of feels to me like Carnegie Hall...." You have to figure out what variables matter in achieving those criteria.
No useful and universal way to define "flat."
*People with long memories will know what kind of demo-proof I am echoing here.
Last edited:
In order to make such a statement, I suspect that this individual has never truly experienced "air" or "quickness" on a music reproduction system.
Well after 3 months I had been back home since and tried this DSS filter on and off. So many times.
But for me, FLAT is the correct response and it sounded most realistic.
I have read SL's slide from the Burning Amp Festival and they make sense. But that is in fact not my experience.
I have to say I'm with JohnK and others. The DSS filter may be Orion-Specific.
My speakers are dipoles as well: S12
What do you mean by your term "flat" that you so emphatically endorse as "realistic"?
Certain knobs on your amp pointing to "0" or electrically flat drive coming out of amp or acoustic testing with a mic or testing with 5 mics or equal-loudness to your ear or "stuff sounds like real" or no obnoxious frequencies or some other meaning?
These different senses of "flat" are the result of differing inputs to your speakers. So they are not the same. Right?
Last edited:
Why the confusion?? I can understand that a speaker may measure different on axis compared to an average in room response but Flat means Flat. No deviation over the measured range. Here is a speaker that measures very flat in room asside from the bass bump primarilly from the measurement technique. I own a pair of them and they sound very "realistic". Graphic from Stereophile.
Rob🙂
Attachments
What I meant by flat is speakers measured at design axis, outdoor, non-gating. 1m from the speakers, 2m from the ground plane.
Incidentally if I pink-noise it indoor 1m from design axis it looks like attached. This is the most realistic sounding even when I know the limitations of those dipoles in the tweeter region.
I do agree however with SL that Stereo System = Illusion Engine. Except that for me, it has to be a transducer, not musical instrument. Hence, flat (as per my definition).
Attachments
Last edited:
The "Spatially Averaged" label at the bottom of the chart may be a very good clue. How many of us do that? How many even know how?
Tho Gainphile seems to have done very well in room with a speaker that measures flat outside.
But this is the crux of the matter, it is hard to define what we mean when declaring a speaker as flat. Virtually any speaker made has a directivity that rises with frequency. Most rooms have absorbtion that rises with frequency. From these two issues it is clear that a speaker/room combo has a time varying spectrum. The early sound and the later sound of a speaker will always have a different spectrum.
So which part do we want flat? For the total reverberent field to be flat (i.e. the steady state response, if beyond the critical distance) then the direct or early sound would have to have a rising response. Is this what we want?
A lot of the papers that I read make a good case that the early sound should be flat and the steady state response should roll downwards with rising frequency. The ear perceives spectral balance with a variable length time window. The window is longer at low frequencies, long enough to encompass the nearby wall reflections. At mid frequencies floor bounces are generally perceived. At high frequencies the window is short enough that only the direct response becomes important.
If we want to talk about "flat speaker response", we need to be very specific about what we mean.
David S.
And with David S's gem of a post and a peek back at the gainphile's OP and early replies, the discussion of "flat" has nowhere to go. The quite meaningful phrase "sounds flat to me" has as much potential for getting final clarification through engineering as the quite meaningful phrase, "I ate till I was full."
But there still remains the practical yet thorny question of "flat" (or better, whether there is a need for changing EQ settings) at different loudness levels. And perhaps adjusting EQ for other listener-dependent time-varying factors such as "adaptation level" (definition: "nice soup, burnt like my mother used to make it").
But there still remains the practical yet thorny question of "flat" (or better, whether there is a need for changing EQ settings) at different loudness levels. And perhaps adjusting EQ for other listener-dependent time-varying factors such as "adaptation level" (definition: "nice soup, burnt like my mother used to make it").
Last edited:
Add to that the weirdos like me who believe that using an omni-directional microphone for in room measurements is quite wrong. (ears don't work like that).
It's a mess, isn't it? But I do like Speaker Dave's last post, for sure.
It's a mess, isn't it? But I do like Speaker Dave's last post, for sure.
Have a look at johnk nao note. That's the flat I want.
About room response etc. Well that's why we only use dipoles. Flat for non uniform speakers will always sound unnatural. Eg. Generic box speakers.
About room response etc. Well that's why we only use dipoles. Flat for non uniform speakers will always sound unnatural. Eg. Generic box speakers.
Having used ESL and other dipoles for the last 40+ years, I am friendly to the concept.
But dipoles do not have a special power to "rise above" the many issues of reflections (early and late), directivity (high freq and low), modes (in all dimensions), etc. Quite the opposite. They "mush together" all those things in a more general and somewhat less salient way. And add more factors (like edge diffraction, rear echo, and rear wave cancellation) into the mix.
Playing fast-and-loose with room modes, etc.... that is their charm.
Does that fit in with anybody's notion of elegant engineering solutions? I don't think so. But works well at home.
Last edited:
Hello David
Just bare with me a minute I may be looking at this from too simplistic a point of view.
If I look at a gated measurement at 1 or 2 meters I am looking at the equivalent of the first arrival “early sound”?? When I measure off axis under the same conditions I am seeing the equivalent of the first reflections before the room works it's own majic on them??. If I had a full set of anechoic measurements they should be approximately the same but without the resolution issues that gating introduces. The reverberant field in a real livingroom would be the sum of the two minus whatever the room adds or subtracts??
I agree with this and run CD set-ups.
How does this effect what we measure either in a gated or an anechoic measurement? I can understand that it may change our target curve but ultimately we don’t measure as we hear in a gated measurement unless we either don’t window or use a sliding window on our measurements that mimics what our ears do. I can only adjust a fixed window in my software. In an anechoic measurement there are ideally no reflections.
We can get a good look at what the on axis early arrival and room artifact free first reflections, polar response, will be. From what I have read they are a good tool as far as predicting how a system will sound in a room.
When we measure we are seeing the “early sound”. By default that’s what we get using gated measurements. The default standard is also On Axis. Based on what we have read and how we take measurements why is there confusion about what we should be looking for or what Flat means?? If I got derailed please get me back on track.
Thanks Rob🙂
Just bare with me a minute I may be looking at this from too simplistic a point of view.
If I look at a gated measurement at 1 or 2 meters I am looking at the equivalent of the first arrival “early sound”?? When I measure off axis under the same conditions I am seeing the equivalent of the first reflections before the room works it's own majic on them??. If I had a full set of anechoic measurements they should be approximately the same but without the resolution issues that gating introduces. The reverberant field in a real livingroom would be the sum of the two minus whatever the room adds or subtracts??
I agree with this and run CD set-ups.
How does this effect what we measure either in a gated or an anechoic measurement? I can understand that it may change our target curve but ultimately we don’t measure as we hear in a gated measurement unless we either don’t window or use a sliding window on our measurements that mimics what our ears do. I can only adjust a fixed window in my software. In an anechoic measurement there are ideally no reflections.
We can get a good look at what the on axis early arrival and room artifact free first reflections, polar response, will be. From what I have read they are a good tool as far as predicting how a system will sound in a room.
When we measure we are seeing the “early sound”. By default that’s what we get using gated measurements. The default standard is also On Axis. Based on what we have read and how we take measurements why is there confusion about what we should be looking for or what Flat means?? If I got derailed please get me back on track.
Thanks Rob🙂
Last edited:
Just to illustrate the above, this is the ungated in-room-response of a 3-way dipole from 200 cm (yellow), 55 cm (white) and 14 cm (green, HT only) distance:
It is easy to see how the high frequencies get attenuated when the room influence becomes dominant.
A dipole like the Nao Note incorporates some very elegant engineering solutions. Technology wise it is way ahead of what you have got at home. Time to extend your knowledge about how dipoles work to todays SOTA. 😉
Rudolf
Attachments
Good questions Rob. Let me see if I can answer them.
What the gated measurement shows depends, of course, on the gating window "on" time. If you gate to the direct sound only, then your measurements should be the same as an anechoic curve. As you know, with gating you need a long window to see the full bass response develop, so we normally ignore it below a certain frequency. It turns out that for those low frequencies it is perceptually more accurate to revert to the steady state (long time window) response.
As for the sum of the two... the reverberent field is really the sum of the direct sound and all reflections. And every reflection has to be viewed as the response from the speaker for that particular radiation angle and the variable absorption of the patch of wall it bounced off of. And later sounds will be from multiple bounces, so they will have a spectrum equal to the speaker at some angle and multiple wall bounce attentuation curves! I know the Harman has done some accurate in-room response simulations. The rest of us would just calculate what the radiated power response was and adjust it with the absorption vs. frequency curve of the room.
I would certainly lean more towards the anechoic curves or a well gated in-room curve as being more close to predicting what the perceived frequency response would be, this is in preference to the in-room steady state (i.e. reverberent field) curves.
Most people that rely on in-room curves learn early on that "flat sounds bright" so they have a pet room target curve. This isn't bad, but in effect they are defining a steady state curve that will, hopefully, give a direct/early sound that is flat. (They may not know it but that is what they are doing.) To do that we have to make assumptions about what the directivity vs. frequency of the typical speaker is, along with the absorption vs. frequency of the average listening room. If we pick an atypical speaker or an oddball room we can equalize to the target curve that worked last time and find that our early response is totally different. The new speaker/room will sound different. Room curves can work but they aren't reliable because we aren't directly measuring the right thing. We aren't measuring the way the ear hears. (Kates wrote a great paper on this some years back.)
So an anechoic curve is better, but it won't tell you anything about how the LF response of the system is strongly impacted by nearby boundaries, or how the floor bounce might impact the midrange. Ideally we would have our variable window and take a measurement in the actual room. At high frequencies it will agree with the anechoic curve. At mid frequencies it has to open up long enough to include at least the floor bounce (see the papers of Soren Bech for confirmation of that), and at low frequencies it can be long (or steady-state) to include all the boundary effects and room standing waves. From what I understand, that would be the ideal measurement and that should be set to flat (okay, with maybe a little bass boost😉).
Regards,
David
Just to illustrate the above, this is the ungated in-room-response of a 3-way dipole from 200 cm (yellow), 55 cm (white) and 14 cm (green, HT only) distance:
Rudolf
Yep, that shows it very well.
And the farther away you get from the speaker or the more the room absorption rolls off at HF the more the top curve will sag. (Anybody ever see the Cinema X curve?)
David
- Status
- Not open for further replies.