Well, that's the real crux of engineering, isn't it? How do you do it at reasonable cost and with what ever other constraints are required?
This setup is very similar to how I had my Yorkville U15s set up, and IMHO can be a successful approach, although not without it's compromises.
The main shortomings are
a) depth. Late/diffuse reflections from the front do seem to influence soundstage depth.
b) spaciousness. this depends a lot on how you treat your first reflection points. In my case, the room was so narrow that I had to put heavy absorption on the first reflection points, making for a very dead room overall, and spaciousness suffered.
Aside from that, my setup was very successful and satisfying - 'imaging' in the sense of left-right placement and speakers disappearing was fantastic. Compared to the best setups I've heard, it was only that last element of 3-D-ness related to the sense of space that was lacking. It did require a fair bit of acoustic treatment though, and the U15's have a very tight 60-degree dispersion pattern which I believe helped.
So, I wouldn't necessarily advocate this as preferable to a more conventional approach, but if it's what you have to deal with it can work.
Yes, of course, but I'm still waiting on the car that's as fast as a Ferrari, quiet as a Roles and priced like a Ford Focus. It's just engineering with some contraints right?
Forget the stuff between the speakers I reckon, in most cases it won't make a whit of difference. For sure, we get the audiophile who puts a blanket on it for critical listening and claims it makes big differences...in most cases rubbish it does.
Because, again in most cases, it is just one of these audiophile 'tricks' he has read about on the net, does it, reports back how good it is. Typical stuff usually associated with cables et al.
Pan the camera shot back of his room and see (again usually) NO treatment, often nice looking glass walls, bare wooden floors (looks nice tho! the wife loves it) etc.
In that morass of early and late reflections, boominess etc etc think a blanket on a tv screen will make a difference??
Note the use of 'usual' and 'mostly'...I am not talking about the proper set ups.
Because, again in most cases, it is just one of these audiophile 'tricks' he has read about on the net, does it, reports back how good it is. Typical stuff usually associated with cables et al.
Pan the camera shot back of his room and see (again usually) NO treatment, often nice looking glass walls, bare wooden floors (looks nice tho! the wife loves it) etc.
In that morass of early and late reflections, boominess etc etc think a blanket on a tv screen will make a difference??
Note the use of 'usual' and 'mostly'...I am not talking about the proper set ups.
Member
Joined 2009
A guide to better imaging
Hi, I'm putting together a small guide and I'd like to post it here for review before making it more public. I'm sorry for repeating topics that have been discussed already.
---------------------------
I would like to present a simple strategy that has helped me achieve better imaging. Equal distances of the left and right speaker to the sidewalls improve the accuracy of the stereo image. If one speaker is closer to the sidewall than the other, a smearing in the image occurs towards that side. When the speakers are properly positioned in the room, the listener can move left and right and the image will shift sideways while losing some clarity.
The underlying concept is not new and has been discussed in numerous publications. A presentation of the basic principles can be found on Singfried Linkwitz's web site. My contribution expands on how one can verify the distance of the left and right acoustic paths by using measuring equipment. I do not understand the underlying psychoacoustic mechanisms that govern this process but the benefits of this approach are readily audible. Everyone is encouraged to try this out and comment on the perceived effects.
Please verify that you meet the following requirements before continuing:
- You have access to measuring equipment.
- Your speakers have an even power response. Omni, Constant Directivity, Fullrange designs as well as some small bookshelf implementations should be the best candidates. Implementations with uneven off-axis response may not be suited for this arrangement.
- Your listening position should bound by two large, acoustically similar, parallel surfaces on your left and right side.(walls)
Good tonal balance takes priority over imaging and is therefore out of the scope of this article. Room modes and early reflections are the primary concern of proper speaker positioning. It's assumed that the reader is aware of their room's limitations. Objects with sharp discontinuities should not be in the way of the direct acoustic path and at least 2ft away from the tweeter and midrange.
We may begin by positioning the measuring microphone close to the preferred listening spot, equally spaced between the left and right wall. Use a tape measure and try to get to about an inch accuracy. In my far from acoustically perfect room there is a huge panoramic window on my left. I understand that the majority of sound reflections on my left will be coming from the window so I treat the glass surface as my left boundary. I have discovered through experimentation that crude acoustic treatments such as blankets and curtains have negligible effects in this application and should not be considered.
Next, using a tape measure let's place the speakers at equal distances to the sidewalls. 2ft or greater clearance is recommended from the side walls. The clearance from the back wall should be at 3ft minimum. Please do not undermine the effect of room modes when selecting the speaker position. When you have the speakers at a satisfactory position run a MLS sweep and have a look at the recorded Impulse Response. For this guide I used a blue plot for the right speaker and red for left.
In the plot the highest amplitude signal is the initial pulse coming directly from the speaker to the microphone. Subsequently at around 1.7ms there's a little bump in the right channel. Judging from the intensity and timing this is most likely a diffraction from furniture in the signal path. Diffractions can be easily identified by covering the suspected sources with a pillow and running another sweep to see if the acoustic signature of the diffraction has changed. Another strategy is to turn the speakers to an angle at which they illuminate the offending object with less intensity while maintaining full energy radiation towards the listening spot.
The second largest disturbance in the plot at 3ms is without doubt the reflection from the side wall. The later reflection at 3.5ms is most likely from the ceiling. Linkwitz recommends that there are no wall reflections for the first 6ms but I ruled that impossible to accomplish with the size of my living room. Yet I get a very sharp stereophonic image which prompts me to believe that although a wider room might be beneficial it may not be necessary.
The most important factor for the focus of the stereophonic image I found to be the equal distance of the speakers to the sidewalls. When centering the speakers one must be careful to align not only the distance to the wall but also the angle at which the speakers face the listening position. It is clear from the initial plot that reflections from the right channel arrive faster than those from the left. In practical terms that translates to the stereophonic image smearing towards the right side. The effects are audibly obvious when one is auditioning image test tones.
Audio measuring equipment can help align the reflections timing very precisely. It can also detect and help neutralize diffractions in the signal path. I think it's important to encourage a culture beyond measuring speaker performance and towards using these tools to properly integrate audio equipment and the listening space.
Hi, I'm putting together a small guide and I'd like to post it here for review before making it more public. I'm sorry for repeating topics that have been discussed already.
---------------------------
I would like to present a simple strategy that has helped me achieve better imaging. Equal distances of the left and right speaker to the sidewalls improve the accuracy of the stereo image. If one speaker is closer to the sidewall than the other, a smearing in the image occurs towards that side. When the speakers are properly positioned in the room, the listener can move left and right and the image will shift sideways while losing some clarity.
The underlying concept is not new and has been discussed in numerous publications. A presentation of the basic principles can be found on Singfried Linkwitz's web site. My contribution expands on how one can verify the distance of the left and right acoustic paths by using measuring equipment. I do not understand the underlying psychoacoustic mechanisms that govern this process but the benefits of this approach are readily audible. Everyone is encouraged to try this out and comment on the perceived effects.
Please verify that you meet the following requirements before continuing:
- You have access to measuring equipment.
- Your speakers have an even power response. Omni, Constant Directivity, Fullrange designs as well as some small bookshelf implementations should be the best candidates. Implementations with uneven off-axis response may not be suited for this arrangement.
- Your listening position should bound by two large, acoustically similar, parallel surfaces on your left and right side.(walls)
Good tonal balance takes priority over imaging and is therefore out of the scope of this article. Room modes and early reflections are the primary concern of proper speaker positioning. It's assumed that the reader is aware of their room's limitations. Objects with sharp discontinuities should not be in the way of the direct acoustic path and at least 2ft away from the tweeter and midrange.
We may begin by positioning the measuring microphone close to the preferred listening spot, equally spaced between the left and right wall. Use a tape measure and try to get to about an inch accuracy. In my far from acoustically perfect room there is a huge panoramic window on my left. I understand that the majority of sound reflections on my left will be coming from the window so I treat the glass surface as my left boundary. I have discovered through experimentation that crude acoustic treatments such as blankets and curtains have negligible effects in this application and should not be considered.
Next, using a tape measure let's place the speakers at equal distances to the sidewalls. 2ft or greater clearance is recommended from the side walls. The clearance from the back wall should be at 3ft minimum. Please do not undermine the effect of room modes when selecting the speaker position. When you have the speakers at a satisfactory position run a MLS sweep and have a look at the recorded Impulse Response. For this guide I used a blue plot for the right speaker and red for left.
In the plot the highest amplitude signal is the initial pulse coming directly from the speaker to the microphone. Subsequently at around 1.7ms there's a little bump in the right channel. Judging from the intensity and timing this is most likely a diffraction from furniture in the signal path. Diffractions can be easily identified by covering the suspected sources with a pillow and running another sweep to see if the acoustic signature of the diffraction has changed. Another strategy is to turn the speakers to an angle at which they illuminate the offending object with less intensity while maintaining full energy radiation towards the listening spot.
The second largest disturbance in the plot at 3ms is without doubt the reflection from the side wall. The later reflection at 3.5ms is most likely from the ceiling. Linkwitz recommends that there are no wall reflections for the first 6ms but I ruled that impossible to accomplish with the size of my living room. Yet I get a very sharp stereophonic image which prompts me to believe that although a wider room might be beneficial it may not be necessary.
The most important factor for the focus of the stereophonic image I found to be the equal distance of the speakers to the sidewalls. When centering the speakers one must be careful to align not only the distance to the wall but also the angle at which the speakers face the listening position. It is clear from the initial plot that reflections from the right channel arrive faster than those from the left. In practical terms that translates to the stereophonic image smearing towards the right side. The effects are audibly obvious when one is auditioning image test tones.
Audio measuring equipment can help align the reflections timing very precisely. It can also detect and help neutralize diffractions in the signal path. I think it's important to encourage a culture beyond measuring speaker performance and towards using these tools to properly integrate audio equipment and the listening space.
This is a great subject. For me, imaging / soundstage is what makes me feel part of the process. Lends the illusion of being there. Without it (poor imaging) the sound is flat, barren and lifeless.
Room treatment is critical to this end, imo. Dealing with first reflection points and creating a RFZ (reflection free zone) around the listener. By RFZ, i mean early reflections. I target those up to 25ms-30ms. By free, and I dont mean -40db in the 0-30ms range, but rather I use the Haas curve for reference.
Referring to this curve, I try to keep the first 20ms reflections to -15db. Between 30ms and 80ms, I aim for reflections in that zone referred to as "spacious"
Using ETC tests, one can get a graph like this:
The green and orange lines I added from the haas data. These illustrate the "inaudable" and "spacious" boundaries, just to help me see visually how things look.
Certainly, what people like in regards to reflections varies. This is just an intro to my approach, and nothing more.
Room treatment is critical to this end, imo. Dealing with first reflection points and creating a RFZ (reflection free zone) around the listener. By RFZ, i mean early reflections. I target those up to 25ms-30ms. By free, and I dont mean -40db in the 0-30ms range, but rather I use the Haas curve for reference.
An externally hosted image should be here but it was not working when we last tested it.
Referring to this curve, I try to keep the first 20ms reflections to -15db. Between 30ms and 80ms, I aim for reflections in that zone referred to as "spacious"
Using ETC tests, one can get a graph like this:
An externally hosted image should be here but it was not working when we last tested it.
The green and orange lines I added from the haas data. These illustrate the "inaudable" and "spacious" boundaries, just to help me see visually how things look.
Certainly, what people like in regards to reflections varies. This is just an intro to my approach, and nothing more.
Last edited:
To me its what is going on below 10 ms that is the most important. The curves are obviously much more complex and things change rapidly in this time region. So I would want to see your green and yellow line and the impulse response (not some time averaged ETC curve) from 0-10 ms. Out to 100 ms is simply a waste of paper. 0-20 ms. maybe (I am assuming that we are talking about a small room like we find in a home where there are lots of reflection < 10 ms.)
I draw it out to 100ms because I am interested in both IMAGING and SOUNDSTAGE. And in my understanding, 20ms-80ms are important to the later and 0-20ms, the former.
The graph I showed was at 1.2k at .33 oct (1122hz-1414hz only). Therefore, it is not representing or trying to represent the entire frequency range. My software allows you to look at different frequency centers with different octave ranges. What i do is separate plots at 317, 500, 800, 1200, 2000, and 3200hz at .33 octave. Since ETC is spectrally blind, I look at several ranges. In looking at a number of plots, breaking the entire frequency range into small pieces, I get a sense for whats happening at different frequencies in regard to different areas of the frequency range.
Last edited:
This is key. I would expand this reasoning to not only considering symmetry to the side walls, but to the entire room from the listening position all the way back to the wall behind the speakers. The left and right should be mirrors of each other as closely as possible. Behind the listening position seems a bit less critical. To what degree I am not sure. But certainly what is in front of you and directly to the side of you, viewing from the listening position, is of great importance.
I understand the concept of "imaging" but I do not know what constitutes "soundstage". (Sounds like just another word for "Spaciuosness").
In a small room I would say that perception is dominated by the very early stuff. Its almost over by 20-30 ms. And not much of our perception is going to change for sound > 50 ms. except for a little additive "spaciousness".
I would not filter or integrate the impulse response and just show it full range over this time period. I might be tempted to high pass at 500-1 kHz (and mybe LP at 10 kHz) since as far as imaging is concerned that is the most critical range of frequencies.
In a small room I would say that perception is dominated by the very early stuff. Its almost over by 20-30 ms. And not much of our perception is going to change for sound > 50 ms. except for a little additive "spaciousness".
I would not filter or integrate the impulse response and just show it full range over this time period. I might be tempted to high pass at 500-1 kHz (and mybe LP at 10 kHz) since as far as imaging is concerned that is the most critical range of frequencies.
I think soundstage is a bit like spaciousness. And I also think that audio reflections >50ms do amount or account for spaciousness. What I want is my cake and eat it too. I want detailed imaging but also a spaciousness or liveliness to the sound.
My logic works like this. This spaciousness has to come from somewhere. That somewhere is the room in the form of reflections. Now if you allow reflections early (<20ms or so), you may get spaciousness, but it would be at the expense of imaging, for the direct sound + those 1st 20ms or so are critical to image detail. So in order to preserve image detail, you have to have those reflections needed for spaciousness arrive later.
Last edited:
You are getting off track. We were talking about what is the right thing to measure NOT what creates image and what creates spaciousness. I never said anything about how to tradeoff image for spaciousness, thats a different topic. Only that if you want to talk about "imaging" then you need better resolution at < 10 ms. and you need to plot the criteria on that plot as it is shown in your plot. What you posted completly ellimiantes what is important for imaging. It will capture spacisouness sure, but not imaging.
I thought soundstage was as relevant to the discussion as imaging given that many people either mean the same thing by both words, confuse their meaning or consider them both equally important.
Looking at the graph I presented, it seems to me the <10ms data is fairly clear. That first vertical divider is about 8ms. The only filter is what I said earlier, a .33 octave response centered at 1200hz measured at the listening position. I have the plots of the other frequencies, but I posted that one for illustration purposes, not to try to convey the entirety of my room response.
In addressing directly what you said. I dont have a saved impulse response. Nor do I have an ETC of just the data to 10ms.
Last edited:
Jim, I think you are on the right track and I like the fact that you are taking a more systematic approach to this when compared to what others are doing. You are to be congratulated.
If you are not already familiar with it, you should look up the concept of the "Haas kicker". Don Davis uses the term in his work (eg, his handbook) although I don't know if the term is original to him. I won't try and define it in 25 words or less on an internet forum, but it touches on some of the issues you are dealing with in terms of "spaciousness" and the positive and negative role of echoes
If you are not already familiar with it, you should look up the concept of the "Haas kicker". Don Davis uses the term in his work (eg, his handbook) although I don't know if the term is original to him. I won't try and define it in 25 words or less on an internet forum, but it touches on some of the issues you are dealing with in terms of "spaciousness" and the positive and negative role of echoes
This is why I objected to the term. If it does not have a clear meaning that one can associate with the literature then of course it is going to be confusing. Its confusing to me. "Imaging" (or source localization) and "spaciousness" are terms that one can find in well regraded texts and papers. "Soundstage" is not such a word. It sounds more audiophile than psychoacoustic.
And I too will agree that what you are doing is a lot better than other approaches that I have seen. I am only trying to suggest improvements in time resolution at the lower times might help to make a better description.
Room Impulse responses are very easy to get with a free piece of software like HolmImpulse, so not having one should not be a limitation.
Actually, I have heard of and at least have a working idea of what it is. For others, a oversimplification of a Haas kicker as I understand it, is basically to re-introduce the signal back into the listening environment in a delayed fashion. Thus avoiding competing with the direct signal along with flexibility as to when and how much to introduce.
In the near future, I intend to work with impulse response more in depth. Right now, my system is completely down while my speakers are getting modified. Once done, I want to synchronize the impulse (timing) of the drivers. That is to say, work on getting the sound from each driver (direct response) to arrive at the ears as closely as possible to the same time.
One idea I am going to try is leaning the speaker. Ok, this requires explanation. Typically, in a 3 way (tweet, mid, woofer) the tweeter output arrives first, then the mid, then the woofer. Now if you have the driver alignment as such whereby the tweet is at the top, the mid in the middle and the woofer at the bottom, you can now imagine that by leaning the speaker back, with the midrange as the pivot point, you can effectively make the tweet further away while at the same time making the woofer closer to you. This, while perhaps not making things perfect, closes the gap between the timing of the tweeter and the woofer which are the farthest apart in terms of timing in typical situations. I realize this isnt a new idea and I am not presenting it as such. But it is something anyone can do, and costs nothing. Even if you dont have software/mics to measure the impulse responses, you can do some simple geometry by taking into account the relative voice coil locations in each driver and trying to get their distance relatively equal relative to the listening position. It may even be worth making the tweet further away from the woofer given a woofers relative slowness in response compared to most tweeters. I dont know how much benefit I will gain by this until I do it. But I think it holds some promise.
One idea I am going to try is leaning the speaker. Ok, this requires explanation. Typically, in a 3 way (tweet, mid, woofer) the tweeter output arrives first, then the mid, then the woofer. Now if you have the driver alignment as such whereby the tweet is at the top, the mid in the middle and the woofer at the bottom, you can now imagine that by leaning the speaker back, with the midrange as the pivot point, you can effectively make the tweet further away while at the same time making the woofer closer to you. This, while perhaps not making things perfect, closes the gap between the timing of the tweeter and the woofer which are the farthest apart in terms of timing in typical situations. I realize this isnt a new idea and I am not presenting it as such. But it is something anyone can do, and costs nothing. Even if you dont have software/mics to measure the impulse responses, you can do some simple geometry by taking into account the relative voice coil locations in each driver and trying to get their distance relatively equal relative to the listening position. It may even be worth making the tweet further away from the woofer given a woofers relative slowness in response compared to most tweeters. I dont know how much benefit I will gain by this until I do it. But I think it holds some promise.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.