I thought it was a lot more simple than that.
If measuring with an RTA in room, then a flat response would actually have boosted treble in reality. This is due to the beaming of the treble frequencies. The bass and midrange would have all their reflections being measured aswell, whereas the tweeter has fewer reflections.
If a speaker is measured flat groundplane outside, then in room it should show a roll of at the treble frequencies.
This info taken from the Master Handbook of Acoustics, If my dodgy memory serves correct.
Rob.
If measuring with an RTA in room, then a flat response would actually have boosted treble in reality. This is due to the beaming of the treble frequencies. The bass and midrange would have all their reflections being measured aswell, whereas the tweeter has fewer reflections.
If a speaker is measured flat groundplane outside, then in room it should show a roll of at the treble frequencies.
This info taken from the Master Handbook of Acoustics, If my dodgy memory serves correct.
Rob.
tubee said:
tubee actually im still working on a living room set up. I havent been able to find out what type of amps are in the JBL LSR 6328P,(oh i wanna just open it to peak!)
other then they are not discrete. But my point was not so mush electronic design but the fact that all amps and speakers sound different, and that in it self is color. The JBL's claim "flat at the mix" so i couldnt see why anyone would want that in their living room.? I may be wrong, i havent done this in a long time but if you were to stand in a room with some flat response speakers and played some pink noise, all freq. equal power, or white noise, wouldn't the average human hear those freq. in 300-8khz range more then others?
impsick said:tubee actually im still working on a living room set up. I havent been able to find out what type of amps are in the JBL LSR 6328P,(oh i wanna just open it to peak!)
The JBL is problably bipolar transistor. When i compare sound of my discrete build surround amp Marantz SR4300 to the hybrid i have the same results: bipolar is sounding some warmer. The winner in warmth/coloured sound was the Quad 303, allmost a warm blanket
impsick said:
Depends on the volume it is playing, with higher levels the whole hearing curve is altered, "more" bass and treble is perceived.
RobWells said:
I think that is the case. I use the deqx and set up the speakers to have a flat response, at a distance of a meter or so (I haven't lugged them outside to do that, I ignore below about 200 hz as I eq the bass flat later anyway).
When taking room measurements from the listening position to do the room eq, there is a natural roll off of the top end due, I presume, to the added distance to the tweeters. In my dumb old days, i had the system 'flat to 20,00' at the listening position. Needless to say it sounded bright and quite brittle. I've since learnt more
I'm very interested in this, with the deqx it is a trivial matter to have the speakers flat on axis, or measure off axis and set THAT flat, indeed the deqx allows one to have both setups simultaneously and be able to flick between them.
Who knows, this may even be 'source dependant', ie each approach may be suitable depending on the recording.
Just finished reading an article by Reg Greene that has some observations pertinant to this discussion, with the added 'bonus' of talking about DSP which is handy for me ha ha.
http://www.regonaudio.com/Audio in Modern Times.pdf
The trick to getting the hotness out of the phantom center image is well known and used often by many mastering engineers. It's a dual channel parametric EQ inserted in M-S (mid-side) matrix. One side of the EQ corrects the center image, the other corrects the sides.
Mastering engineers can EQ lead vocals for example without affecting background vocals that are mixed off to the sides.
Many mastering processors use this process, but call it something like Spectral Imager (TC Electronics Finalizer) or some other voodoo sounding name.
The same process can be used to EQ the playback chain. If you use a computer to play back your music and have the ability to insert DSP plugins in your playback chain, there is freebie M/S encoder/decoder plug from Voxengo:
http://www.voxengo.com/product/msed/
You would insert one instance at the beginning of the chain- for encoding, then you would insert a dual channel EQ (with separate controls for left and right- not ganged), then you would insert second instance of the msed for decoding. Adjustments in the left channel of the EQ would affect the center image only, adjustments in the right channel would affect the stereo imaged sounds.
So a slight dip at 7kHz and 2.5kHz in the Mid channel, compensated by a corresponding slight boost in the Side channel would keep the overall power response constant and make flat voiced speakers sound right.
Here is an example of what the processing would look like in Console
Mastering engineers can EQ lead vocals for example without affecting background vocals that are mixed off to the sides.
Many mastering processors use this process, but call it something like Spectral Imager (TC Electronics Finalizer) or some other voodoo sounding name.
The same process can be used to EQ the playback chain. If you use a computer to play back your music and have the ability to insert DSP plugins in your playback chain, there is freebie M/S encoder/decoder plug from Voxengo:
http://www.voxengo.com/product/msed/
You would insert one instance at the beginning of the chain- for encoding, then you would insert a dual channel EQ (with separate controls for left and right- not ganged), then you would insert second instance of the msed for decoding. Adjustments in the left channel of the EQ would affect the center image only, adjustments in the right channel would affect the stereo imaged sounds.
So a slight dip at 7kHz and 2.5kHz in the Mid channel, compensated by a corresponding slight boost in the Side channel would keep the overall power response constant and make flat voiced speakers sound right.
Here is an example of what the processing would look like in Console
An externally hosted image should be here but it was not working when we last tested it.
Thunau said:
Jan: Thank you! I'll see if this works in foobar or via my Julia card. Its odd that this has never crept its way to the speaker design world.
To really have a full grasp on speaker design, you need not just the technical chops and a good ear, but the knowledge found in recording studios, so that you can voice in a controlled, intelligent manner. Thanks for the post, this type of info is gold.
Dave
DDF,
My speakers were autoEQ to flat, and I added some lowerend through the DEQ and lowered 3.1khz down 2db with a Q of 3. The speakers were toed in at around 15 degrees in an almost equilateral setup.
I liked it how it was before, but the voicing was still unnatural and still foward sounding. I do like it a bit better by going with what you suggested in the original post. I still need to play with the toeing a bit more to see if there's any improvements.
Anything else you'd suggest for my speakers?
Regards,
Kev
My speakers were autoEQ to flat, and I added some lowerend through the DEQ and lowered 3.1khz down 2db with a Q of 3. The speakers were toed in at around 15 degrees in an almost equilateral setup.
I liked it how it was before, but the voicing was still unnatural and still foward sounding. I do like it a bit better by going with what you suggested in the original post. I still need to play with the toeing a bit more to see if there's any improvements.
Anything else you'd suggest for my speakers?
Regards,
Kev
Balance is a difficult things. It all depends on how the overall system is done. You mensioned traditional 2-way construction which is pretty straight forward in this aspect.
Have you measured your speakers off axis? Since flat on axis doesn't tell that much. Power response would and the best approximation you can have is to measure a box speaker from 60 degrees horizontal off axis. That shows pretty much the whole picture how the speaker radiates in your room. To all directions. No worth thinking that you're listening on axis or some small off axis response since you don't. Sound that you hear is a mixture of all reflection within certain timewindow.
You mensioned that you need to pad down 7Khz and such to get it sound right. So I'd estimate you've crossed your speaker 2,5-3,5Khz with steep filter, didn't use waveguide to match the radiation patterns with large midwoofer and caused bump in the power response just above the crossingpoint. Midwoofer starts to turn directive, off axis response drops and driver beams in large angles. Then it's crossed with steep filter to wide radiating tweeter causing addition to overall radiation just around that 5-8Khz range. This is pretty normal effect in many DIY and commercial speakers.
So there is nothing wrong with flat on axis response. Basically with box speakers it should be quite important even but you need to fix that power response better using different methods. What's the correct power response is more difficult subject and there is no one answer to it. In larger space with woodwalls (that pass bass) power response should propably roll off more towards the treble than in small concrete room. But overally slight power response roll off towards treble is a must because most (+90%) recordings are tailored to speakers that don't radiate all frequences with even amounts (=flat power response).
Even power response is the key here. Measure your speakers 60 degrees off axis and check it out. Waveguided tweeter is one solution, gentler crossover slope another. Gentle slope adds tweeters wider radiation over wider frequencyrange so that dramatical directivity changes don't appear. Naturally this also leads into overall wider radiating speaker.
You can also pad down the problemarea but that doesn't solve the problem. It just drops off few dB from direct sound which is raised back in there with wide off axis dispersion. Less direct sound to the listening spot, more reflected sound which leads to lower accuracy.
Jussi
Have you measured your speakers off axis? Since flat on axis doesn't tell that much. Power response would and the best approximation you can have is to measure a box speaker from 60 degrees horizontal off axis. That shows pretty much the whole picture how the speaker radiates in your room. To all directions. No worth thinking that you're listening on axis or some small off axis response since you don't. Sound that you hear is a mixture of all reflection within certain timewindow.
You mensioned that you need to pad down 7Khz and such to get it sound right. So I'd estimate you've crossed your speaker 2,5-3,5Khz with steep filter, didn't use waveguide to match the radiation patterns with large midwoofer and caused bump in the power response just above the crossingpoint. Midwoofer starts to turn directive, off axis response drops and driver beams in large angles. Then it's crossed with steep filter to wide radiating tweeter causing addition to overall radiation just around that 5-8Khz range. This is pretty normal effect in many DIY and commercial speakers.
So there is nothing wrong with flat on axis response. Basically with box speakers it should be quite important even but you need to fix that power response better using different methods. What's the correct power response is more difficult subject and there is no one answer to it. In larger space with woodwalls (that pass bass) power response should propably roll off more towards the treble than in small concrete room. But overally slight power response roll off towards treble is a must because most (+90%) recordings are tailored to speakers that don't radiate all frequences with even amounts (=flat power response).
Even power response is the key here. Measure your speakers 60 degrees off axis and check it out. Waveguided tweeter is one solution, gentler crossover slope another. Gentle slope adds tweeters wider radiation over wider frequencyrange so that dramatical directivity changes don't appear. Naturally this also leads into overall wider radiating speaker.
You can also pad down the problemarea but that doesn't solve the problem. It just drops off few dB from direct sound which is raised back in there with wide off axis dispersion. Less direct sound to the listening spot, more reflected sound which leads to lower accuracy.
Jussi
DDF said:
Try these varying references:
http://www.linkwitzlab.com/models.htm#H
http://www.harbeth.co.uk/faq/index.php#13
http://www.speakerbuilder.net/web_files/Archives/122702.htm
http://www.axiomaudio.com/boards/sh...1&Main=98599&Words=+Twebbz&topic=&Search=true
From these and other posts the reasoning behind it is either because (from Linkwitz):
"Around 3 kHz our hearing is less sensitive to diffuse fields. Recording microphones, though, are usually flat in frequency response even under diffuse field conditions. When such recordings are played back over loudspeakers, there is more energy in the 3 kHz region than we would have perceived if present at the recording venue and a degree of unnaturalness is introduced."
or to reduce the sibilance in the playback.
DDF said:
AJ: Here, you’re repeating an argument I used to make on the MAD board. I guess I had that coming.
Hi Dave,
I was most likely there on Mad, but I don't recall this particular subject. Actually, Mad memories seem to be receding into the distance, unfortunately. I think Jan (who was probably there also) answered it to my satisfaction. I see nothing wrong with a de-emphasis filter - as long as it is not fixed.
The last part about the kids I can't argue with, although we all have our priorities. Re-mastering a 1000+ cds to the HD wouldn't be fun
cheers,
AJ
hi guys
looks like I'll be able to give a listening test about this soon, will make it rather simple, just make the response flat about (suppose) 30 degrees off axis rather than on axis, don't think that exactly mimics what Dave was talking about but at least will test having a flat 'power function' I think it is called.
Have been wondering about all this recently, not sure if it exactly fits in here, but it does kinda fit in with the thread title heh heh.
It came up earlier in the thread. A flat on-axis speaker placed in the 'average' room....I reckon in most cases it will have an in-room response that falls away in the higher frequencies, due I suppose to the differential absorption characteristics of the room on the varying frequencies (feel free to correct my assumptions as I go).
As a logical further step, I feel therefore that if a speaker(s) was setup to measure flat at the listening position, it would (by the inverse of the above principal-if true that is) as a result NOT measure flat on-axis in an anechoic chamber, in other words it would be bright by measurement and I'm pretty certain it would sound bright too.
Now, I've done a bit of a search on the net yet any hard evidence of this seems thin on the ground.
Maybe it is not as clear cut as I thought??
I can kinda draw those conclusions from some of the stuff Mr Toole says, but it's not too explicitly stated. He makes the case that a flat on axis response (with good off axis characteristics as well) seems to be universally preferred in listening tests, and then has a room response algorithm they use to predict the in-room response, which generally shows that those flat speakers measure with a fall towards high when placed in a room.
The closest I've come is Reg Greene (Regonaudio). For a start, he is one of the few places i have found that includes measurements in the reviews! including (in many cases) an in-room measurement as well. So I have found a few examples where a flat on axis speaker falls towards the highs in an in-room measurement (some didn't by the way).
However, he came pretty close to stating what I started with above
""""The next graph shows the 1/6th RTA at the listening position. This is actually a very smooth curve for in-room response, and the smooth downward slope from deep bass to 10k is what is usually regarded as ideal.""""
The graph didn't post here but thats OK.
On another review he came as close as I've seen anywhere
""""Gradient Revolution--In-room response
I have inserted in the graph a "target curve", a gently down sloping curve that is of the type that DSP room correction designers have discovered is needed for the room response if the speaker is to sound natural. (Actually the need for this down sloping had been observed much earlier ,notably by Ortofon in their work on what measurements actually were relevant to the sound of a speaker in a room. Many others had also noted this in the analog -EQ era. It is universally agreed upon, very nearly, that some such curve needs to be used.) """"
Once again the graph didn't paste in here, but the ''target curve'' he drew that represented the almost 'ideal' in room response sloped from the bass (in this case around 25 hz) to be down roughly 4db @ 6.5k, then steeper after that to what looks like an extra 6-7 db@20k.
Soooo, to cut a long post short, is it the case that this sort of 'psycho-acoustic' research is in actual fact thin on the ground, or is it more the case that I in my ignorance do not know where to look??
Hope I didn't bore you (too much!!), but does anyone have any links, or thoughts, that go to this??
thanks a lot
looks like I'll be able to give a listening test about this soon, will make it rather simple, just make the response flat about (suppose) 30 degrees off axis rather than on axis, don't think that exactly mimics what Dave was talking about but at least will test having a flat 'power function' I think it is called.
Have been wondering about all this recently, not sure if it exactly fits in here, but it does kinda fit in with the thread title heh heh.
It came up earlier in the thread. A flat on-axis speaker placed in the 'average' room....I reckon in most cases it will have an in-room response that falls away in the higher frequencies, due I suppose to the differential absorption characteristics of the room on the varying frequencies (feel free to correct my assumptions as I go).
As a logical further step, I feel therefore that if a speaker(s) was setup to measure flat at the listening position, it would (by the inverse of the above principal-if true that is) as a result NOT measure flat on-axis in an anechoic chamber, in other words it would be bright by measurement and I'm pretty certain it would sound bright too.
Now, I've done a bit of a search on the net yet any hard evidence of this seems thin on the ground.
Maybe it is not as clear cut as I thought??
I can kinda draw those conclusions from some of the stuff Mr Toole says, but it's not too explicitly stated. He makes the case that a flat on axis response (with good off axis characteristics as well) seems to be universally preferred in listening tests, and then has a room response algorithm they use to predict the in-room response, which generally shows that those flat speakers measure with a fall towards high when placed in a room.
The closest I've come is Reg Greene (Regonaudio). For a start, he is one of the few places i have found that includes measurements in the reviews! including (in many cases) an in-room measurement as well. So I have found a few examples where a flat on axis speaker falls towards the highs in an in-room measurement (some didn't by the way).
However, he came pretty close to stating what I started with above
""""The next graph shows the 1/6th RTA at the listening position. This is actually a very smooth curve for in-room response, and the smooth downward slope from deep bass to 10k is what is usually regarded as ideal.""""
The graph didn't post here but thats OK.
On another review he came as close as I've seen anywhere
""""Gradient Revolution--In-room response
I have inserted in the graph a "target curve", a gently down sloping curve that is of the type that DSP room correction designers have discovered is needed for the room response if the speaker is to sound natural. (Actually the need for this down sloping had been observed much earlier ,notably by Ortofon in their work on what measurements actually were relevant to the sound of a speaker in a room. Many others had also noted this in the analog -EQ era. It is universally agreed upon, very nearly, that some such curve needs to be used.) """"
Once again the graph didn't paste in here, but the ''target curve'' he drew that represented the almost 'ideal' in room response sloped from the bass (in this case around 25 hz) to be down roughly 4db @ 6.5k, then steeper after that to what looks like an extra 6-7 db@20k.
Soooo, to cut a long post short, is it the case that this sort of 'psycho-acoustic' research is in actual fact thin on the ground, or is it more the case that I in my ignorance do not know where to look??
Hope I didn't bore you (too much!!), but does anyone have any links, or thoughts, that go to this??
thanks a lot
FWIW, I remember that years ago the Large Advent had a tweeter level toggle switch on the back, and the preferred position was slightly rolled off. They made some claim that with the very best source material, you could use it in the flat position. Methinks they knew what response sounded best, but wanted a fallback so the speakers would look better on paper in a nearfield test. Back when I was doing diy speakers, before a lot of the technical points that are common knowledge today, were common, I spent a lot of time tweeking for flat response. I very clearly remember achieving ruler flat response at 1m, and thinking the sound was about the worst I'd ever heard. I also remember taking what I thought were well balanced and decent speakers into a showroom to compare with others. Mine sounded dismal, as everything else was boosted and bright, so it would stand out and sell to the general public. Though I don't know the full technical basis, IMO nearfield flat is far from desirable.
Two things strike me in the posts above and certainly agree with my experience.
Yep, that's how you sell stuff. Same for televisions and computer monitors. They are far from "flat" in the showroom. They tend to have the blue boosted way up, as well as the contrast and brightness pushed thru the roof. It helps catch your eye, just like the boosted treble and high bass catch your ear in the show room.
Get that stuff home and it will be fatiguing in the long run -even in the short run. A well calibrated video monitor is a thing of beauty, but its too subtle to "pop off the shelf" at the general public. Same for speakers.
As for the falling response, it does happen in nature and our ears and brain expect it. When I used to mix live sound there would be bands who wanted "the guy who mixed our CD" to mix the concert. It always sounded awful. These guys were good audio techs, I'd heard the CDs. But the live mix was always far too bright.
Studio engineers sit 2 or 3 feet from control room monitors. For sure they hear every little detail and high frequency. Try that sound in a large venue and it's an ear splitter - very screechy. It sounds all wrong. A good live engineer knows this and doesn't try to push the top end, he lets it roll off in accordance with what sounds natural in the size of the venue.
On the flip side, Reg Green talks about the evolution of classical instruments, even reworking old ones to fit modern concert halls. The new instruments are brighter than in centuries past because the highs die out in the large modern halls. So there has been a treble boost in instruments to carry in the larger halls that attenuate the high end. (Is that why close miked violin sounds so awful?)
Yep, that's how you sell stuff. Same for televisions and computer monitors. They are far from "flat" in the showroom. They tend to have the blue boosted way up, as well as the contrast and brightness pushed thru the roof. It helps catch your eye, just like the boosted treble and high bass catch your ear in the show room.
Get that stuff home and it will be fatiguing in the long run -even in the short run. A well calibrated video monitor is a thing of beauty, but its too subtle to "pop off the shelf" at the general public. Same for speakers.
As for the falling response, it does happen in nature and our ears and brain expect it. When I used to mix live sound there would be bands who wanted "the guy who mixed our CD" to mix the concert. It always sounded awful.
These guys were good audio techs, I'd heard the CDs. But the live mix was always far too bright.Studio engineers sit 2 or 3 feet from control room monitors. For sure they hear every little detail and high frequency. Try that sound in a large venue and it's an ear splitter - very screechy. It sounds all wrong. A good live engineer knows this and doesn't try to push the top end, he lets it roll off in accordance with what sounds natural in the size of the venue.
On the flip side, Reg Green talks about the evolution of classical instruments, even reworking old ones to fit modern concert halls. The new instruments are brighter than in centuries past because the highs die out in the large modern halls. So there has been a treble boost in instruments to carry in the larger halls that attenuate the high end. (Is that why close miked violin sounds so awful?)
Thanks for the replies.
I know it's 'early days' yet, but so far it seems that 'hard evidence' is indeed not readily available, and that maybe my almost ineffectual attempts to find data of this sort mirrors the paucity of information??
It seems logical to me, but often logic is a poor guide to the way things really are.
Has no-one got any links to information that covers this area??
Posts have to be shorter now since I broke my wrist y/day, typing has become slow and laborious!
I know it's 'early days' yet, but so far it seems that 'hard evidence' is indeed not readily available, and that maybe my almost ineffectual attempts to find data of this sort mirrors the paucity of information??
It seems logical to me, but often logic is a poor guide to the way things really are.
Has no-one got any links to information that covers this area??
Posts have to be shorter now since I broke my wrist y/day, typing has become slow and laborious!
yeah Trev
maybe because the question has appeared a couple of pages into a thread means that not as many people are seeing it??, or maybe I have asked the question poorly.
Or indeed, the 'definitive' answer is not out there, after all there are many competing philosophies when it comes to audio if you know what I mean.
I feel I can make a logical and reasonably compelling argument based on the very few references I've come across, but as i said earlier those arguments are more inferred than stated.
Stereophile for example posted a composite picture of (say) 50 or so systems that measured flat on axis, but unfortunately did not post a corresponding composite of the in-room response of all those speakers. That at least would have given some sort of insight of the relative FR at the listening position.
I guess another advantage of posting in the DIY section is that very many here (obviously) measure their own speakers, and (I assume) most set it up to be flat on-axis (and if not, why not??) and possibly a good percentage of them would also have in room (at listening posn) measurements as well. If, as I strongly suspect, those in room measurements do consistently show a falling response in the higher frequencies then it would tend to support the idea that a flat response at the listening position would measure 'hot' on axis and also sound bright in the room.
Will wait a little and see if any links appear here, if not perhaps i should repost the topic fresh????
Have I posted in the wrong place???
I still thank those who responded, but am more and more beginning to think it's off topic, if so I apologise.
maybe because the question has appeared a couple of pages into a thread means that not as many people are seeing it??, or maybe I have asked the question poorly.
Or indeed, the 'definitive' answer is not out there, after all there are many competing philosophies when it comes to audio if you know what I mean.
I feel I can make a logical and reasonably compelling argument based on the very few references I've come across, but as i said earlier those arguments are more inferred than stated.
Stereophile for example posted a composite picture of (say) 50 or so systems that measured flat on axis, but unfortunately did not post a corresponding composite of the in-room response of all those speakers. That at least would have given some sort of insight of the relative FR at the listening position.
I guess another advantage of posting in the DIY section is that very many here (obviously) measure their own speakers, and (I assume) most set it up to be flat on-axis (and if not, why not??) and possibly a good percentage of them would also have in room (at listening posn) measurements as well. If, as I strongly suspect, those in room measurements do consistently show a falling response in the higher frequencies then it would tend to support the idea that a flat response at the listening position would measure 'hot' on axis and also sound bright in the room.
Will wait a little and see if any links appear here, if not perhaps i should repost the topic fresh????
Have I posted in the wrong place???
I still thank those who responded, but am more and more beginning to think it's off topic, if so I apologise.
Hi Terry J,
Get yourself a copy of The Master Handbook of Acoustics .
It's a great read and there is stuff in there about RTA's and room curves etc.
Cheers,
Rob.
Get yourself a copy of The Master Handbook of Acoustics .
It's a great read and there is stuff in there about RTA's and room curves etc.
Cheers,
Rob.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.