Yes, I realize my thoughts may not be clear, it's a hunch I have. I should not have quoted you...
I need to think this over
Last edited:
Frequency response is everything!?...
- Thread starter audiofull
- Start date
Yes, I realize my thoughts may not be clear, it's a hunch I have. I should not have quoted you...
I need to think this over
I think you are on to a valid point here. There are ways that a system can fail in terms of measured fidelity that are similar enough to what happens in live venues that they don't sound bad to us at all. I've been learning this recently as I attempt to fix various measured "problems" in my room. One was to fix a fairly deep notch in my speaker's response at 80Hz. I found that doing so made it seem too loud at that frequency. So, the dip as measured will stay. My ears say it's not a dip. Another was a rear wall reflection causing an immediate cancel effect between 100 and 200 Hz. I fixed that with absorbers but perceptually it made little difference. Those frequencies just arrive a little later from side wall reflections, and my brain seems to fuse it back together so the tone still sounds natural.
A general rule of thumb with equalizing is don't try to fix nulls. EQ down some peaks, if they sound bad, just enough to perceptually sound natural, and generally avoid narrow, High Q filters.
Last edited:
A safe way to avoid high Q filters, is by applying psycho acoustic smoothing to the measurement or use frequency dependent window. Cancellation notches are best left alone if you are using IIR filters. FIR filters are a much stronger tool, so sometimes notches can be fixed to some degree without a penalty.
Yes, I've usually applied psycho acoustic smoothing to any measurement, so for a long time I never noticed the notch as an actual speaker response problem. I recently started playing with FIR filters after they became available to me in a software update. The FIR can completely eliminate that notch, but it comes at the expense of some significant latency, which can be a problem with movies and video games. It also sounds strange. That 80Hz seems really powerful, visceral, and it turns out that it's highly reverberant in a way the FIR filters cannot calm down. I think the cause is a standing wave that forms inside the bass horn folds. It looks like a peak inside the horn's mouth.
The notch is most likely room induced and not from the horn itself. Look at the group delay in the measurement, you should see a peak at 80-85 hz, probably around 50-100 ms. Standing waves produces frequency peaks accompanied with ringing / decay, not a notch. This is why FIR filters are so much powerful than IIR filter, you can compensate for both frequency amplitude variations and correct for timing issues (phase) at the same time.
Tim, did you try the DRC Designer, I'm quite sure you will like the result.
Tim, did you try the DRC Designer, I'm quite sure you will like the result.
I'm convinced it is induced by the horn itself. I've seen it in 3 different rooms now, and have tested it in various positions in those rooms. It's astounding to put your head in the horn and hear loud bass, and then walk all over the room and barely be able to hear it at all, anywhere. I also did some extensive testing of the volume in the back chamber, the throat, and at various points along the horn's length. It's loud in the back chamber, loud at the throat and loud at the mouth. The air is just getting pushed back and forth between those two points, causing not much energy at all to escape the horn. I've been able to simulate this effect so I know it's real and possible. The sound can get stuck inside the horn. Too small a mouth for the length of the horn is a big issue. It might be just that rather than the folds; the horn mouths are just too narrow. One possible solution might be some controlled venting of the back chamber.
Yes, there's a sharp peak in group delay to 60ms at the notch frequency. The reason I think it's the horn itself, and its mouth being too narrow for its length, is that the notch frequency corresponds to the length of the horn path at 1/2 wavelength. When I set that up in a wave simulator the sound gets stuck inside the horn. It's pretty amazing to see the super loud sound inside the horn and practically nothing beyond the horn mouth at that frequency. If I can extend the mouth further the simulator shows that the problem is solved. I could do that with some extensive cabinetry on either side of the the horn. That turns these into increasingly huge speakers, but they could be multi-purpose with storage areas on either side. I could do this without impinging on the space much, and if done tastefully might even look good.
Tim that is a wild phenomena.. I once did a lecture theatre and we used perforated sheetrock for rear wall.. when tested it gave this massive reflection for vocals .. acoustic engineer did calcs and discovered the ratio of dia to depth of perforations gave perfect reflections in vocal range ! It so hard to process when you looked at this massive wall full of holes.
We didnt use those guys again !
So this is like transmission line being1/4 length of fs
Do you have a rough cross section of horn .. no worries if not
I am really enjoying this deep dive into acoustics. Thanks to all participating
Phil
Tim
I recently tried the ground plane method for measuring in room. Put something solid like a 1x1 foot ply panel in listening spot and point mic down at say 80% and 1/2" from panel ( pointing towards speakers. This removes floor reflections and early reflection of speaker bits from the equation and gives a what seems good picture of what is happening in bass and lower mid.
I used a rect window and no time restrictions.
It has helped with bass /mid xover and to see the other room dips and peaks.
Supposedly you get a 6db boost as your capture direct plus immediate reflection. Apart from off axis roll off the tweeter was not far off either
Low fr in room is so hard to measure and this is another data point.
Apologies if you are already all over it
Phil
I recently tried the ground plane method for measuring in room. Put something solid like a 1x1 foot ply panel in listening spot and point mic down at say 80% and 1/2" from panel ( pointing towards speakers. This removes floor reflections and early reflection of speaker bits from the equation and gives a what seems good picture of what is happening in bass and lower mid.
I used a rect window and no time restrictions.
It has helped with bass /mid xover and to see the other room dips and peaks.
Supposedly you get a 6db boost as your capture direct plus immediate reflection. Apart from off axis roll off the tweeter was not far off either
Low fr in room is so hard to measure and this is another data point.
Apologies if you are already all over it
Phil
I do have a cross section drawing somewhere that I'll have to dig up, but for now I can just explain it as basically an exponential flare that has a constant height of about 18", and widens from 4" at the throat to 2' at the mouth over a distance of about 6 feet. At the mouth there is about a 2' roundover on each side so it gets maybe another foot or so of depth from that.
If I were to take these to a large flat area outside I could get a good ground plane measurement. At over 7' tall I don't think the microphone would need to be all that close to the ground. It would be interesting to see what they measure in a situation like that. There's a school nearby that even has an inside corner on part of the building that faces into a large open playground area. That would let me test the horns corner loaded into a free field. Moving the cabinets out there would be a major undertaking. With some hand trucks and a small crew of helpers I could get it done. Maybe after I finish painting the house and replacing the carpet I'll think about that.
The idea of the ground plane is that you cut out floor reflections . This works really well outside with no other boundaries
I do it in a large driveway so you get a really long time period before any reflections and get accurate clean plots right down to 20hz (assuming your not in a flight path )
It is a big undertaking and I try to do it once during construction.
I was pleasantly surprised how clean a bass plot I achieved using it inside where bass is an almighty mess to plot
Yes, if you're outside you can get rid of every other reflection, but unless you have a tall tower, the ground is hard to get away from. I don't think I have much in the way of floor to ceiling reflections in my room, but the wall reflections are a mess. If you have a single woofer you can get a long ways by just putting the mic. right up to the woofer. When you've got a whole bunch of woofers in horns it's effectively impossible to get close enough to them as a group and meaningfully reduce the effect of room reflections. If you're close to one woofer there are still other woofers that you're many feet from.
I
No need to hike stuff into the air.
Plus you are measuring at a distance where waves and interactions are fully developed.
You can use a rectangular window and long time period .. it was a revelation!
If you point all drivers to the mic on the ground you get one mirror ground reflection only at that point so its pure.. just +6db. Langstrom Holland who I bought my clio system from explained it.
No need to hike stuff into the air.
Plus you are measuring at a distance where waves and interactions are fully developed.
You can use a rectangular window and long time period .. it was a revelation!
As an acoustical engineer, I would argue that relative phase is a more important measurement over frequency response. Both acoustical and electrical phase anomalies are more perceptual to sound quality attributes, effecting amplitude, spatiality, and timbre. Fairly poor frequency response can sound good or bad, but poor relative phase never sounds good.
Is it the case that poor relative phase will cause problems with frequency response .. or am I misunderstanding ?
I certainly agree that sharp dips in fr aren't too bad but long shallow dips and peaks seem to make a big difference to me
Phil
Similar threads
