After playing with the position of the speakers. I came to the conclusion speakers pointing at my ears didn't sound good.
I like to place the speakers with a small angle to the center. It sounds more open with bigger angle the interference effects becomes audible.
It means I listen with a angle of 26 degrees to the center to the driver, this means also will mean a decrease of spl it will be near flat when it has on axes a few dB increase.
I like to place the speakers with a small angle to the center. It sounds more open with bigger angle the interference effects becomes audible.
It means I listen with a angle of 26 degrees to the center to the driver, this means also will mean a decrease of spl it will be near flat when it has on axes a few dB increase.
Last edited:
Thank you for putting it so succinctly. That has been my experience too.
Is there a way to measure the direct sound when one is so far away?
Regards
Mike
Any of the impulse or FFT type measuring systems (Holm, MLSSA) will let you gate the impulse response to cut off before room reflections arrive. (Read the DIY Holm thread, recommended)
David S.
I agree, this has been our general experience as well. Dave has certainly given an excellent explanation.
Best Regards,
TerryO
A lack of general consensuses on this issue is not surprising. A reproduction process starting with a uniquely tailored virtual event (recording) doesn't help. Then this virtual event finds it's way out into the world. The event is attempted to be reassembled via an incredible variety of complicated equipment, in god knows what combination's, playing back at different levels, in different environments and so on. I think it would be a safe bet to assume that no two playback conditions are, or for that mater, could be the same. Let alone considerations for the human interpretive aspects.
What ever you do, you can not exactly duplicate the original event. Seems silly to write something so obvious. Know that for arguments sake one may use perfect playback as a goal. Got to have some kind of base line. Determining what compromises can be made with minimal damage to the original event seems to me to be the most productive way to approach speaker design. A designer can not choose the environment their product will work in. If fact, the designer has little if any influence on how their product will be used. They must make a few broad assumptions and run with that. Not exactly the best situation!
So we get back to choosing a few basic performance parameters, making decisions on what performance criteria can be sacrificed with the least harm at a given price point. And frequency response is usually where the design compromises start. The compromises employed are well known. Once in a great while, some one breaks new ground and finds a way of preserving a bit more accuracy at a particular price point. An evolutionary design process found in all manufactured goods. Not very elegant, however, quite practical.
After frequency response one would look at harmonic, linear and IM distortion products, power response and dynamics. And as I said before, once you get the frequency response part where you need it, the rest become much easier. This assumes of course the drivers used are inherently low distortion designs suitable for the intended play back levels.
What ever you do, you can not exactly duplicate the original event. Seems silly to write something so obvious. Know that for arguments sake one may use perfect playback as a goal. Got to have some kind of base line. Determining what compromises can be made with minimal damage to the original event seems to me to be the most productive way to approach speaker design. A designer can not choose the environment their product will work in. If fact, the designer has little if any influence on how their product will be used. They must make a few broad assumptions and run with that. Not exactly the best situation!
So we get back to choosing a few basic performance parameters, making decisions on what performance criteria can be sacrificed with the least harm at a given price point. And frequency response is usually where the design compromises start. The compromises employed are well known. Once in a great while, some one breaks new ground and finds a way of preserving a bit more accuracy at a particular price point. An evolutionary design process found in all manufactured goods. Not very elegant, however, quite practical.
After frequency response one would look at harmonic, linear and IM distortion products, power response and dynamics. And as I said before, once you get the frequency response part where you need it, the rest become much easier. This assumes of course the drivers used are inherently low distortion designs suitable for the intended play back levels.
That's effectively it. You set your min and max window sizes and then it uses shorter windows at higher frequency bins, transitioning to longer windows at the lower frequency bin. With boundary distances in play in home listening, the intent is to try and get a better representation of how near boundaries influence perceived loudness, over frequency: less at high frequencies, more at low.
Its useful but very hard to determine what the "right" min and max are to arrive at a curve with an accurate representation of the tonal balance in room. It's of course room size and boundary proximity dependent, but even within that context, the number of variables in play are too extreme to arrive at a generalized min/max window size setting.
Here's an old B&K idealized in room response:
That looks about right, but I think this is going to vary depending on room size.
As far as nonlinear distortion is concerned--unless it's gross, the ear masks it. You can train yourself to hear it if you'd like--see Klippel--by why?
Dan
That looks about right, but I think this is going to vary depending on room size.
As far as nonlinear distortion is concerned--unless it's gross, the ear masks it. You can train yourself to hear it if you'd like--see Klippel--by why?
Dan
Last edited:
Hi David,
- The room does virtually not alter the direct sound
- the room alters the reflections and is eating up highs
So where do you want to put the engery to keep the direct sound flat ? It is flat already and where should that energy come from that makes too bright if the room is eating it especially in a cinema or any other dead room ?
My theory as to why highs are being rolled of is a little different: When tweeters start to beam, and they do for sure, you will have decreasing reflections and increasing direct sound. And I believe that could be one reason for too bright. That is less reflected sound (less sound power) and rising DI => excess direct sound in the highs compared to the rest of the speaker.
Another data point for my previous post is that many headphones (one example is my Sennheiser HD 650) have a falling FR although there are no reflections at play. This is all direct sound and it seems too bright for our perception. I know that the incident angles to the ear may be different with headphones compared to stereo sound from speakers but the main point is the absence of reflections vs. perception.
Oliver
That is understood and agreed. I am not advocating equalizing power to flat in the listening room, hell no ! 🙂 When I talk about power response in this context I mean measured under anechoic conditions, just like Toole describes and uses it.
OK, so far.
So let's assume flat on axis and flat sound power under anechoic conditions. And didn't you say above that
- The room does virtually not alter the direct sound
- the room alters the reflections and is eating up highs
So where do you want to put the engery to keep the direct sound flat ? It is flat already and where should that energy come from that makes too bright if the room is eating it especially in a cinema or any other dead room ?
My theory as to why highs are being rolled of is a little different: When tweeters start to beam, and they do for sure, you will have decreasing reflections and increasing direct sound. And I believe that could be one reason for too bright. That is less reflected sound (less sound power) and rising DI => excess direct sound in the highs compared to the rest of the speaker.
Another data point for my previous post is that many headphones (one example is my Sennheiser HD 650) have a falling FR although there are no reflections at play. This is all direct sound and it seems too bright for our perception. I know that the incident angles to the ear may be different with headphones compared to stereo sound from speakers but the main point is the absence of reflections vs. perception.
Oliver
I think you are saying that flat direct sound and flat power response, anechoically measured, should be good. The Lipshitz and Vanderkooy test I referred to disagrees with that. They weren't equalizing to a flat reverberent field but to a flat power radiation from the speaker and found it sounded too bright. It appears tht what sounds natural is a flat direct response and considerably falling reverberent field. Note that this would be totally natural for our experiences with live instruments. Take a violin from room to room to room of many different sizes. The natural experience is that the larger the room the longer the sound sustains (longer RT) and the duller the later sound becomes.
We don't need to add energy to make the direct sound flat. It can start that way by design and the room will not influence it. The room only influences the reverberent spectrum.
Highs have a higher proportion of direct sound but not a higher level. That is, the direct sound curve can (and should) be flat. The system only sounds too bright when we begin to flatten the reverberent field.
Headphones are a different case where, depending on whether you are measuring at the ear canal or in the ear canal you will need very different and non-flat target curves. I don't know that a lack of reflections makes the sound inherently bright?
What is show in the diagram the response or the damping of a room.Here's an old B&K idealized in room response:
That looks about right, but I think this is going to vary depending on room size.
As far as nonlinear distortion is concerned--unless it's gross, the ear masks it. You can train yourself to hear it if you'd like--see Klippel--by why?
Dan
When it is the response the high frequency is already damped -6dB.And then they want to damp it even 10 dB more can not be and certainly not what is right to my ears.
Let me back up again. I believe there are some terminology and/or language issues that become difficult to unravel.
One of those terms is "flat": What is flat ? Straight and strictly horizontal (I believe this is meant) ? Straight but tilted up or down with no humps and bumps (aka strictly monotonic in/decreasing) ?
Quote from Toole about the sound power and how it is measured:
"Sound power is intended to represent all the sounds arriving at the listening position."
=> ("late") reflections.
"Thus, the on-axis curve has a very low weighting because it is in the middle of other, closely adjacent measurement points, and measurements further off-axis have a higher weighting because of the larger surface area that is represented by each of those measurements."
So we have an on axis FR curve representing the direct sound (the listening window curve however also represents a part of that if people are not directly in the sweet spot) and we have a curve for sound power describing the reclections that come back via the room. This curve will naturally fall because of absorption and beaming.
Would speakers placed in the garden or in a anechoic chamber as extremes sound naturally ? There is normally no reverberant field.
What sounds natural to me (in terms of stereo), is a speaker (in a certain room) (now quote from Toole, page 459) "...with two qualities: wide dispersion, thereby promoting higher levels of reflected sound, and a relatively constant directivity index so that the direct-sound and reflected-sound-curves have a similar shape."
And that clearly requires a more constant sound power curve than anything with a rising and high DI. Ultimately, the beaming will prevent constant sound power/flat DI all the way up.
That is my current belief.
One of those terms is "flat": What is flat ? Straight and strictly horizontal (I believe this is meant) ? Straight but tilted up or down with no humps and bumps (aka strictly monotonic in/decreasing) ?
Quote from Toole about the sound power and how it is measured:
"Sound power is intended to represent all the sounds arriving at the listening position."
=> ("late") reflections.
"Thus, the on-axis curve has a very low weighting because it is in the middle of other, closely adjacent measurement points, and measurements further off-axis have a higher weighting because of the larger surface area that is represented by each of those measurements."
So we have an on axis FR curve representing the direct sound (the listening window curve however also represents a part of that if people are not directly in the sweet spot) and we have a curve for sound power describing the reclections that come back via the room. This curve will naturally fall because of absorption and beaming.
I disagree with that and I believe there are other reasons as to why we perceive something as too bright. Thus my example with the falling headphones FR, which represents only direct sound. They had to slope the direct sound down to make it sound good.
That equals a rising and finally high DI.
Would speakers placed in the garden or in a anechoic chamber as extremes sound naturally ? There is normally no reverberant field.
No, more sound radiated off axis by the tweer(s) = wider dispersion, less beaming.
What sounds natural to me (in terms of stereo), is a speaker (in a certain room) (now quote from Toole, page 459) "...with two qualities: wide dispersion, thereby promoting higher levels of reflected sound, and a relatively constant directivity index so that the direct-sound and reflected-sound-curves have a similar shape."
And that clearly requires a more constant sound power curve than anything with a rising and high DI. Ultimately, the beaming will prevent constant sound power/flat DI all the way up.
That is my current belief.
Yes, and the larger the room and greater the distance between source and listener, the lower the direct response may be in the total steady state response measurements. (Measurements, not perception.)
Yes, natural for a dead environment. Clearly when listening in a dead environment, out of doors, or an anechoic chamber, the power response is immaterial and only the direct response can be heard. It should be flat. Even Acoustic Research who were advocates of rolled off response to emulate a "concert hall balance" commented that they had to flatten the direct response of systems when recording speakers in anechoic chambers for live vs. recorded tests.
Again, this is why the L&V tests were interesting. They tried achieving flat power radiation both ways: by having a conventional beaming speaker with EQ to give brighter on-axis response, and by adding power to the reverberent field alone, leaving the on-axis response flat. In both cases the system sounded too bright. Same with the Omnisat design: wider dispersion than usual meant the system needed even the on-axis response tilted down for a natural balance. The end conclusion is that flat on-axis response with a moderate and rising directivity is what works.
Toole likes to sit on the fence regarding power response but his curves in the group tests are fairly definitive. His trials show a wide range of power response shape can rank high in listening tests (agreeing with L&V) as long as the directivity generally rises and is free from resonances. If power response is non critical then directivity index shape is equally noncritical. Having wide dispersion is good for a natural stereo image but not a requirement for a flat perceived balance. Read what he says about the Quad ESL63. Due to its higher than average d.i. it is downgraded significantly in "spaciousness", more so in mono tests than stereo tests. Too high a directivity can be bad for stereo (envelopment) but not necessarily bad for perceived frequency balance.
David S.
Dave, thanks for the exceedingly clear explanation again!
It's the idealized response of the speaker in the room. I'm not sure what you are talking about in the second part.
Dan
It's the idealized response of the speaker in the room. I'm not sure what you are talking about in the second part.
Dan
Being a live sound front of house guy for many years, I have heard a number of studio guys try to mix live. Always much, much too bright. Some of them would rip your ears off. These guys are not hacks, they just are not used to mixing live. So they rarely got the tonal balance right (among other mistakes).
I think it's because they are so used to nearfield. But in a huge room, it should not sound that way. Our brains don't expect it to. It may have something to do with our ears, too. But bright in a big room does not sound right. And flat in a medium size room does not either.
So... what is the response of the room and system in the typical mastering suit? Flat at the listening position? Probably not. And if it isn't and yours is, how is that going to sound?
I think it's because they are so used to nearfield. But in a huge room, it should not sound that way. Our brains don't expect it to. It may have something to do with our ears, too. But bright in a big room does not sound right. And flat in a medium size room does not either.
So... what is the response of the room and system in the typical mastering suit? Flat at the listening position? Probably not. And if it isn't and yours is, how is that going to sound?
David, could you please provide a ref. or a link to the L&V paper ? I couldn't find it in the thread.
First it is being stated that power response and DI are insignificant and later the power response is being blamed for making sound too bright. This is indeed exceedingly clear and very convincing 😀
First it is being stated that power response and DI are insignificant and later the power response is being blamed for making sound too bright. This is indeed exceedingly clear and very convincing 😀
AES E-Library: Experiments in Direct/Reverberant Ratio Modification
Never said "insignificant", always that power response must not be equalized to flat.
Also look at:
AES E-Library: A Perceptual Criterion for Loudspeaker Evaluation
AES E-Library: The Influence of the Room and of Loudspeaker Position on the Timbre of Reproduced Sound in Domestic Rooms
AES E-Library: Perception of Reproduced Sound: Audibility of Individual Reflections in a Complete Sound Field, II
AES E-Library: A New, Psychoacoustically More Correct Way of Measuring Loudspeaker Frequency Responses
All the answers are in those papers.
David S.
Never said "insignificant", always that power response must not be equalized to flat.
Also look at:
AES E-Library: A Perceptual Criterion for Loudspeaker Evaluation
AES E-Library: The Influence of the Room and of Loudspeaker Position on the Timbre of Reproduced Sound in Domestic Rooms
AES E-Library: Perception of Reproduced Sound: Audibility of Individual Reflections in a Complete Sound Field, II
AES E-Library: A New, Psychoacoustically More Correct Way of Measuring Loudspeaker Frequency Responses
All the answers are in those papers.
David S.
The cinema guys seem to have a better handle on this. Not only is there an "X" curve for theater EQ, but they have evolved it into a room size related curve. Large theaters, small theaters, and mastering suites all have specified curves with various HF rolloffs.
Now, if they had a psychoacoustically correct way to measure they wouldn't need to do this, but it is better than nothing.
David
Hi David,
thanks for the links. Haven't read it yet but I'm always willing to learn.
I did not say that you said...but it relates to 4/#34. It was thinking out loud in the context of this thread.
Oliver
thanks for the links. Haven't read it yet but I'm always willing to learn.
I did not say that you said...but it relates to 4/#34. It was thinking out loud in the context of this thread.
Oliver
- Status
- Not open for further replies.