Getting back for a moment to the difference between headphone and speaker based reflection simulation (per my reservations with Kates method), Bech found two interesting outcomes (see AES E-Library: Browse Entire Database
- reflections from the front wall are more easily detected free field than in headphone sims (as used by Kates). Free field thresholds are 5 to 10 dB lower. So, for the threshold of front wall refection to affect timbre, take the results I posted for Kates, and subtract ~ 5 dB from the threshold taken at 10ms and lower, and subtract ~ 10dB for the thresholds at delays above 10ms.
- reflections simulated free field at 30 degree to 60 degree lateral (ie the first side wall reflection) have thresholds similar to teh headphone listening.
So based on Bech's test, roughly, side wall reflections have to be ~ 5 to 10 dB higher before affecting timbre for side wall compared to front wall (and not too far a stretch, floor).
Dave Dal Farra
PS:
In a related search, found a paper very applicable to this discussion:
http://orbit.dtu.dk/getResource?recordId=264155&objectId=1&versionId=1
- reflections from the front wall are more easily detected free field than in headphone sims (as used by Kates). Free field thresholds are 5 to 10 dB lower. So, for the threshold of front wall refection to affect timbre, take the results I posted for Kates, and subtract ~ 5 dB from the threshold taken at 10ms and lower, and subtract ~ 10dB for the thresholds at delays above 10ms.
- reflections simulated free field at 30 degree to 60 degree lateral (ie the first side wall reflection) have thresholds similar to teh headphone listening.
So based on Bech's test, roughly, side wall reflections have to be ~ 5 to 10 dB higher before affecting timbre for side wall compared to front wall (and not too far a stretch, floor).
Dave Dal Farra
PS:
In a related search, found a paper very applicable to this discussion:
http://orbit.dtu.dk/getResource?recordId=264155&objectId=1&versionId=1
I think this is the Toole/Olive paper DDF brought up earlier:
http://www.aes.org/tmpFiles/elib/20110319/6079.pdf
If anyone wants to read it.
Dan
http://www.aes.org/tmpFiles/elib/20110319/6079.pdf
If anyone wants to read it.
Dan
It's not intuitively obvious, but it's what Moulton found. Interestingly enough, he may be in lonely company, see my follow up post on the audibility of side wall reflection from Bech.
I can't say why their results differ from each other, the papers never provide enough experimental set up information to allow that level of rigour in review. But I tend to believe Bech on this one for singular reflections. However, Moulton's tests may have had a much higher density of reflections (ie real life!) so its too early to judge.
The key here is if the signal envelope presented in the inner ear changes due to these reflections. If it does, good chance it's audible. For this, a non-linear mechanical model through the middle ear, oval window and basilar membrane would be helpful. They exist (wrote one in grad school) but never see them used.
This is an area of research calling out for an auditory codec based analysis but there's no money in that and as you know.
Dave Dal Farra
Its always been the baffle edge diffraction that concerns me, because it is very early and not at all insignificant. We are talking abouyt "reflections", but diffraction is in this same time regime. I have found that elliminating all very early reflections AND diffractions (from nearby objects, etc.) in the vicinity of the loudspeakers has a pronounced effect on the image.
When I was at Knowles we had something like this running in near real time and we used it to simulate hearing loss, so it is certainly doable.
But in a very real sense isn't what we are trying to do is to "drill down" through this nonlinear model to find the essence of what is important at the level of the actual room and loudspeaker. This is how I always viewed the subject topic - what is it that we are actually sensitive to in the room/loudspeaker -> ear interface.
When you look at it this way its obvious that a simple one microphone frequency response at a point has very little chance of yielding anything that is going to correlate to perception. Its just not that simple.
In my opinion any strong single reflection that makes our brain go back into "localization mode" is potentially detrimental to the spatial attributes of a recording. Removing reflections from the listening room does help with localization and low level details without taking away spaciousness and depth. The key is the rooms reflection pattern.
Obviously others experience "depth" despite a low direct-reflection-ratio, e.g. with omnis. Lots of reflections might prevent "localization mode" too.
I think you're right about the timbre change introduced by an early side wall reflection. For the last year my listening room has been relatively small at 4.8 x 3.45 metres, and moderately live at a bit above 0.4s for RT30.
For most of that time I've had the speakers firing "across" the 3.45 metre axis of the room - eg the speakers are against the long wall, keeping them about 1.3 metres from the side walls. Even though they're quite close to the wall behind them (about 0.7 metres from the front of the speaker) the tonal balance was quite good and largely insensitive to moving the speakers sideways.
Even though the room is fairly live the listening position at 2.4 metres was close enough to put me just within the "direct field" (direct>reverberant) and the perceived balance was largely that of the on axis response of the speaker.
I decided to swap the room around a few days ago so the speakers were on the short wall firing down the length of the room. My first reaction when I tried playing it was YUCK. Tonal balance was considerably out (woolly sounding and lacking in presence) and with the new listening distance of 3 metres and the speakers fairly near the side walls it's clear I'm now well into the reverberant field, and I perceive largely the timbre of the power response - I can move the tweeter back and forward which in near field on axis response makes a huge change, but at the far listening distance there is almost no change in perceived balance doing this...
Due to the narrow width of the room in this configuration the furtherest I can get the centre of the speakers away from the side walls is about 0.7 metres, upon trying to optimize the sound a bit with speaker positioning I very quickly found that the distance from the side wall was extremely fussy - just moving the speakers a few inches toward or away from the side wall is making dramatic changes to the overall timbrel balance, and it was very difficult to get a satisfactory balance.
I can only assume it's due to the comb filtering from the side wall, combined with the off axis response of the speaker in the direction where the reflection is occurring...
Years ago in other houses I always had my speakers on the short wall firing along the long length, and was sceptical of putting speakers against the long wall, but that was in a much larger room that allowed me to keep the speakers further from the side walls, and the listening position was also well forward into the room, putting me in the direct field.
I'm not so sceptical now though, and after I've finished measuring and playing with the current configuration I'll be changing the room back around to how it was 😛
One difference with baffle diffraction from the reflections from nearby room objects and side walls though, is that it's at least consistent, and therefore part of the speaker design.
A speaker will have a certain baffle diffraction signature, of which the amplitude portion of it will be compensated for to some extent in the network (on axis anyway) but making moderate positional changes of the speaker in the room wont alter the diffraction signature of the speaker appreciably. (Except for changing it's height off the floor)
On the other hand early reflections from side walls (as I posted about in my previous reply) and reflection/refraction from nearby objects such as TV's will affect the response in a way that is very critical to speaker placement, and thus becomes an "unknown" from the speaker designers perspective.
Email notifications
AFAIK, you have 3 choices when subscribing to a thread.
You should not get more than 1 instant, until you visit the thread again. As was said above, this is to keep your email from being flooded and to keep the load on the email server low.
The forum is really set up to be read online, not as a mail list, tho a fair number of folks do use it that way.
AFAIK, you have 3 choices when subscribing to a thread.
- Instant Notification
- Daily Digest
- Weekly Digest
You should not get more than 1 instant, until you visit the thread again. As was said above, this is to keep your email from being flooded and to keep the load on the email server low.
The forum is really set up to be read online, not as a mail list, tho a fair number of folks do use it that way.
Creating such a model allows simulation to be carried out over extensive sets and varieties of conditions. Changes in the inner ear spectrum are then assessed to set priorities for DBT. Focus on DBT new avenues showing large innear ear spectrum changes.
This technique is important as it allows the affects of delay and sound onset direction to be independently and exhaustively explored.
In the past and present, the research plods along using DBT to assess what to DBT next.
This thread has been interesting, illustrating numerous examples and causes why a single mic measurement doesn't "tell it as it is". Hopefully a few new converts will be made, thereby allowing less diversions in the discussion.
Dave Dal Farra
Just want to say thanks for the above paper link. I too get lost in many of the AES papers, where mathmeticians use acronyms and obscure citations and such regularly. It's as if they're talking to themselves, rather than others. I like plain English. 🙂
I have to agree with the possibility that diffraction effects in the first 1ms and perhaps as much as 2 or 3ms are the most concerning because the distance between our ears is 6" (or 0.5ms) and therefore we should worry about the possible affects on our brains perception of stereo imaging and localization.
But in my mind, it's the delayed horizontal reflections between 2-6ms that are still debatable.... As mentioned by another poster, Moulton and a few others seem to agree that the lateral reflections are actually quite useful in helping our brain to correctly perceive the stereo image and prevent the "inside your head" effect of stereo reproduction.
Perhaps there is not so much of a negative affect from these 2-6ms reflects from the sidewalls (assuming a relatively smooth off axis response and perhaps some diffusion of course).
In the end, I'm not yet convinced that the 2-6ms reflections are always so detrimental as many believe... If we eliminate all the early reflections would it not sound like an anechoic chamber which everyone seems to agree actually sounds quite bad?
Maybe Moulton has some good ideas?
My room is just slightly larger than with RT30 near .3s. With my dipoles located at 70cm from the sidewall, the side reflections occur at a very high incident angle to the front baffle of the speaker and thus I don't detect any adverse affects from the sidewall reflection at all.
You assume comb filtering but maybe, just maybe, the off axis response of your speakers is contributing to your sensitivity with regard to placement near the side wall. Were you listening to dynamic monopoles by chance?
Last edited:
I believe that Moulton is fairly alone in this regard. I do not think that 2 - 6 ms reflections do any "good" whatever. There is no "inside the head" if they are not there as long as there is cross-talk from a free field situation.
Yes they're dynamic monopoles. They're somewhat directional from about 2Khz up though, as they 8" full rangers on a pretty wide baffle crossed to a wave guide loaded ribbon tweeter at 4Khz. (The FR driver narrows to about +/- 30 degrees at the crossover frequency, with the ribbon ~ +/-45 degrees horizontally over most of it's range)
Since first swapping the room around I've done a bit more investigating and seem to have nailed down the various effects and got a relatively satisfactory result, although still not as good as I had with the other room configuration.
After trying different things it's become apparent that the side wall distance is subjectively affecting primarily the upper bass, lower midrange region from about 150-400Hz, in particular certain distances are putting a hole in that region making it sound too bright and thin, with other distances boosting that region making it sound woolly.
There doesn't seem to be nearly as much effect in the upper midrange or treble as I first thought, hardly any in fact. This is consistent with essentially getting a "floor bounce" effect from the side wall, not surprising when the midrange driver is 76cm off the floor, and the side wall is roughly within the same distance range from the driver.
If the cancellation frequency from a side wall reflection happens to be the same as a cancellation frequency from a floor bounce reflection then the combined effect could be quite severe - perhaps that's what I'm noticing as I adjust the distance, and find a critical spot where it sounds particularly bad - two notches are lining up with each other.
By moving the speakers in away from the wall as much as possible whilst constrained by other objects (right speaker 0.7m, left speaker 0.9m) and adjusting the EQ from 160-400Hz down by -1.5dB I'm now getting a subjectively good balance across that range.
I still wasn't happy with the upper midrange or imaging though, and all along I noticed that it was mainly the right speaker that seemed very indistinct and unfocused - the left one wasn't too bad, which got me thinking about room asymmetry.
The left wall had a sofa along it, plus I have some sound absorbent material in what is now the rear left corner, however apart from a shallow fake fire place what is now the right wall is largely bare. Also the middle to right hand corner of the rear wall immediately behind the sofa is also bare.
That whole right rear corner that my right ear was pointing directly into was as bare and live as you can get.
On a hunch I took a small bean bag (1m long, 30cm diameter) and stuffed it up onto the edge of the sofa right in the corner of the room - more or less at ear level. I was hoping to diffuse and absorb the sound entering the corner.
WELL! What a difference. 😱 Of all the other changes I made, this one simple change made the most improvement by far.
Immediately the perceived asymmetric balance in both timbre and image localization between left and right speaker was cured, even though the right speaker is still 0.2m closer to the wall than the left speaker. The first 2 metres of wall between the right speaker and fireplace (the early reflection zone) is still bare and untreated as well - I haven't done anything to that area, and yet the side wall doesn't seem to be having any significant detrimental effect on the imaging. (After the 150-400Hz range was taken care of anyway)
The perceived lack of presence and imaging from the right speaker is gone, and tonal balance from midrange through treble now sounds correct, and more or less the same as it was with the speakers set up pointing across the room instead of along it.
So what's happened ? My tentative conclusion thus far is - the speakers have enough directivity in the upper midrange and treble (especially when toed in) that the early side wall reflection is not significantly shifting the perceived tonal balance from upper midrange up, and the time delay is probably also (just) sufficient to avoid the reflection being "combined" by the brain, instead getting treated as a delayed arrival.
However in the lower midrange area where the speakers aren't very directional we're getting a floor bounce like effect from the side wall that is affecting the response in that region and is fairly sensitive to side wall distance, especially in that the response variations may coincide or not with the actual floor bounce depending on relative distance.
Getting the speakers as far from the side walls as I can and applying a small amount of EQ in that frequency range seems to be enough to tame this aspect of the problem.
As for the bean bag and the upper frequencies - unlike the near side walls the location of the live rear right room corner is within the on axis response arc of both speakers at high frequencies, so both speakers will be fully illuminating that corner with high frequencies.
I've always felt that the ear is unduly sensitive to reverberation coming from the sides - possibly due to HRTF, (although I don't want to open that can of worms again after its been thrashed to death earlier in the thread 😉 ) and the corner of the room is about 100 degrees from the listener axis, and very close to the listener since the listening position is at the rear wall.
Perhaps cutting down the level of these late reflections coming in from near the sides or just behind you is very important in preventing them from overpowering the direct signal and early reflections.
The change in overall RT30 adding the bean bag was only from 0.4 to 0.38 however the reduction in reverberation coming from that corner directly into the right ear is probably quite dramatic.
Another change I noticed when adding the bean bag which I was not expecting, is an improvement in the tightness of the bass. Another problem with the room around this way (and the limited amount of positional adjustment I have open to me with the speakers due to objects) is that I'm ending up with a rather large hole (>10dB) in the bass from 60-80Hz, which gives the bass an odd and unsatisfying "hollow sound".
I noticed immediately after adding the bean bag the bass was noticeably improved - both in tightness and a slight increase in mid bass. I thought I must be imagining it so I took some before and after measurements with ARTA, and indeed the depressed 60-80Hz midbass region increased by about 1dB with the bean bag in place, also with a 1dB reduction in a troublesome narrow peak at 90Hz.
I can understand why it would change, but I would have thought a single bean bag of that size would not have any significant effect on bass frequencies... has anyone else tried placing large bean bags right in the corners and measured their effect on bass ? 🙂
Last edited:
now that was really funny! thank You!
I'd say that you are well into the region of "fine tuning" now. 🙂 Not just the overall room tone, but where the reflections actually come from. The better things get, the more you'll notice changes like that - as you are well aware.
Can't say that I have, but it's common to put rolls of fiberglass in the corners to control bass. It works. Are your bean bags actually beans, or Styrofoam pellets? Come to think of it, I've put bags of shipping peanuts in the corners but never noticed a difference.
- Status
- Not open for further replies.