• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

Keith_W DSP system

My theory is that the extra 5 Hz extension is loosening up the house to the point I'm loosing both modes and potential cabin gain.
I'd say you need to go quite a bit below that to provoke the structure of your house. My subs are quite big and numerous so they reach down to 6-7Hz to v high levels ok and it really needs them driven hard that low to make my house croak. Obviously ymmv but that sort of thing (going really loud well below 15Hz to make a house shake) seems common in typical US construction from what I've seen/read (unless you're going to genuinely absurd SPL in which case all bets are off)
 
(unless you're going to genuinely absurd SPL in which case all bets are off)
Guilty as charged ! :D

My experience is anything above 30Hz, that is loud enough will shake any non brick, non concrete, construction.....and there's always the roof and windows etc...and floor trusses if above ground. So I guess I'm saying it's more about loud than how low. (once a certain f-3 is reached)

But that's a bit in contrast with my theory that dropping down an additional 5Hz, is accelerating my house breakup.
Only reason I'm even entertaining the theory, is that a long series of DIY subs...sealed, then 4 different double 18" ported builds, have all used the same drivers.
The latest version is simply larger and is ported lower than previous. Currently running three double 18"s, using bms18n862'.s
 
I've taken time to go through your entire thread the last few days @Keith_W and it's been a great read. I've been doing similar things for the same amount of time but from the Mac side of DSP and only with REW and rePhase. If anything, the thread has got me real interested in getting Acourate in the future to see how much further I can go with what I've got.

I too, am dealing with interesting speakers but in a rather inflexible space and I'm not as concerned with perfection at a main LP as broad coverage through the room. Home built 3-way active XO speakers flying from the ceiling with 2 subs in opposite diagonal corners of a 16x20x11' room. I've made a lot of the same mistakes you have over the years and still make stupid mistakes/experiments all the time!

Currently I'm using a mix of IIR and FIR to get my sound tuned. I prefer a modular approach as mentioned above by others but all hosted within Mitchco's HLC host as AU or VST "modules". Incredible flexibility to plunk in a 2ch FIR for S/B/M or T and test it out before it gets baked into the XO. Here's the filters I'm currently using. Looking at the impulse, the subs FIR give the appearance of potential pre-ring but it's not evident in listening. At this time I'm only using correction up to 125Hz.

You've inspired me to create my own "journey" thread like this, maybe this winter when I have more time though.

N3B.jpg
 

Attachments

  • LREQ N3B.mdat.pdf
    648 KB · Views: 53
You've inspired me to create my own "journey" thread like this, maybe this winter when I have more time though.

Thank you! One motivation for creating this thread is that it is good for learning. Just look at the eager response from ASR members when they realized that I might have screwed up my compression measurement ;) Every time I post something new I feel as if I am throwing meat to the alligators. Yeah they are brutal. But they are also well-meaning. I have to go into this and keep my ego in check because pride inhibits learning. Be prepared to swallow it and admit you are wrong, and you will learn much faster. So I would like to see you start your own thread, just be prepared for some ASR tough love.
 
LOL, that's why I put up my .mdat (just remove the .pdf extension) - can't hide anything there! I've no illusions about my skills and would happily take constructive critique to make things better. But I don't want to hijack your thread.

Like I said, reading through the whole thread it was fun seeing you try things. Some things were all new to me, some things I was like :facepalm: but the results and dialogue as to "why" with others chiming with their knowledge is invaluable. For someone like me who's trying to integrate 8 drivers without an app to guide me through it's all gold.

I'm glad I didn't get a mic before I ripped out the XO on my speakers 15 years ago, I'm sure my passives were atrocious! But then I started the journey with a miniDSP4x10 and have never looked back. Finding ASR 6 years ago has done more to improve things than anything. Perhaps we should start a thread with just active XO people to share our methods and ask questions - if anyone's gone this route they obviously like to tinker and try things.
 
My experience is anything above 30Hz, that is loud enough will shake any non brick, non concrete, construction.....and there's always the roof and windows etc...and floor trusses if above ground. So I guess I'm saying it's more about loud than how low. (once a certain f-3 is reached)
I have to really really crank it to make the structure complain where "really crank it" is in the region of cinema reference level (so 120dB+ in the subs), even 110dB down to 6-7Hz for me is still clean. To verify, I went back and fixed my gain problem and ran a few sweeps from ~85 to ~110dB and it's a pretty neat linear increase. The woofers oddly show an excess gain of about 0.5dB in the 150-250Hz region from 85 to 110dB & it's clearly the woofers as it's only apparent in the band they operate in, not sure what would create that effect though it's pretty small so not really a practical concern (can't say I'm fully focused on precise details at that sort of SPL!). I can certainly imagine a different construction would behave v differently though.
 
Thank you! One motivation for creating this thread is that it is good for learning. Just look at the eager response from ASR members when they realized that I might have screwed up my compression measurement ;) Every time I post something new I feel as if I am throwing meat to the alligators. Yeah they are brutal. But they are also well-meaning. I have to go into this and keep my ego in check because pride inhibits learning. Be prepared to swallow it and admit you are wrong, and you will learn much faster. So I would like to see you start your own thread, just be prepared for some ASR tough love.
Hi Keith, may I first say how much i enjoy your posts on ASR. You bring a great deal on energy and experience that is often in areas/ways that not available to us all, or that we haven't yet explored. Thx man!
And pls know if/when it's me that comes on too direct, that I'm indeed well meaning..as you graciously surmise.
Just a crusty old codger I am ;) Or so my beautiful younger missus keeps telling me !!
 
I have to really really crank it to make the structure complain where "really crank it" is in the region of cinema reference level (so 120dB+ in the subs), even 110dB down to 6-7Hz for me is still clean. To verify, I went back and fixed my gain problem and ran a few sweeps from ~85 to ~110dB and it's a pretty neat linear increase. The woofers oddly show an excess gain of about 0.5dB in the 150-250Hz region from 85 to 110dB & it's clearly the woofers as it's only apparent in the band they operate in, not sure what would create that effect though it's pretty small so not really a practical concern (can't say I'm fully focused on precise details at that sort of SPL!). I can certainly imagine a different construction would behave v differently though.
Very cool. Thx. Sounds like you have a well dialed in room response. I'd call the tiny bit of woofer gain non-existent...within the range of measurement to measurement error.
Congrats.
I had a room like that in my previous residence. Was built from ground up with audio in mind. Dimensions per acoustic formulas for optimizing modal dispersion. Slab on grade, brick, super stout...other than roof ended up being weak link when i put corner stacked Labhorn subs into the room Lol. Would blur you damn vision ! I do miss that room.
 
For quite a while I have wanted to try a weird experiment - what happens if I use an extra pair of speakers, digitally delay them, then point it at the wall to create "ambience"? The idea is similar to digital reverb ("concert hall effect" on AVR's) - but those don't work because the reverb is coming out from the same speakers which is producing the direct sound. Linkwitz said that we need "enveloping" sound, so the idea was to take the extra speakers and place them behind me in both corners of the room, pointing away from the listening position. But I did not have enough DAC channels to do it. That's when I found out that the answer has been under my nose all along - VB-Audio Matrix!

I downloaded it earlier today and got it to work.

1722172784541.png


First step: grab my computer speakers (AudioEngine A2) and linearize them.

1722173091108.png


Before linearization (red) and after linearization (green). Even before linearization it measured quite OK. There is not much <120Hz, but that's no problem - I don't need them to make bass. Look at that textbook step response!

1722173203975.png


Because the AudioEngines are on a different interface, I suspected that the latency would be off, so I checked with Acourate. I came up with a method to sweep both the tweeter on my main system and the little AudioEngine at the same time. It involved:

- Create a 2 way XO for the Audioengine. I used a 50Hz corner frequency. Then I deleted XO1 and renamed XO2 to XO1.
- Copy a previously generated XO for my mains tweeter and renamed it XO2. So now I have XO1 = Audioengine high passed at 50Hz, and XO2 = tweeter reference from main speakers.
- Set Acourate to output to VB Audio Matrix. Route all the inputs to the correct outputs.
- Generate a Multiway WAV and logsweep both. Result as above.

I can see that the AudioEngine is 9010 samples (187.5ms) delayed with respect to the tweeters. I then time aligned both speakers.

1722172591667.png


Then I assigned all my DAC's to ASIO channels and enabled the virtual ASIO I/O channel.

1722172658740.png


Next step is to route all the channels appropriately. Look at that, I have 3 DAC's in the same ASIO channel! VB Matrix allows me to delay the AudioEngine, so I dialled in a 20ms delay.

1722172700702.png


Then I set up Acourate Convolver to send channels 1-8 to the main speakers, and channels 9-10 to the "ambience" speakers.

1722173620220.png


Then I had fun placing the "ambience" speakers in various positions in the room. My main speakers have a little gap at the back, perfect to hide a pair of small computer speakers, effectively converting it into an "almost" dipole. I am limited by cable length as to where I can place the speakers - the furthest I can place them apart is on either front corner of the room. Can't place them where I want just yet.

The effect is pretty remarkable - the 20ms delay was chosen because it lays outside the Haas fusion zone and early enough to be perceived as "ambience" but not so late that it is perceived as an "echo". It really does add spaciousness, and the effect is far more convincing than "concert hall" reverb.
 
Last edited:
For quite a while I have wanted to try a weird experiment - what happens if I use an extra pair of speakers, digitally delay them, then point it at the wall to create "ambience"? The idea is similar to digital reverb ("concert hall effect" on AVR's) - but those don't work because the reverb is coming out from the same speakers which is producing the direct sound. Linkwitz said that we need "enveloping" sound, so the idea was to take the extra speakers and place them behind me in both corners of the room, pointing away from the listening position. But I did not have enough DAC channels to do it. That's when I found out that the answer has been under my nose all along - VB-Audio Matrix!

I downloaded it earlier today and got it to work.

View attachment 383450

First step: grab my computer speakers (AudioEngine A2) and linearize them.

View attachment 383452

Before linearization (red) and after linearization (green). Even before linearization it measured quite OK. There is not much <120Hz, but that's no problem - I don't need them to make bass. Look at that textbook step response!

View attachment 383453

Because the AudioEngines are on a different interface, I suspected that the latency would be off, so I checked with Acourate. I came up with a method to sweep both the tweeter on my main system and the little AudioEngine at the same time. It involved:

- Create a 2 way XO for the Audioengine. I used a 50Hz corner frequency. Then I deleted XO1 and renamed XO2 to XO1.
- Copy a previously generated XO for my mains tweeter and renamed it XO2. So now I have XO1 = Audioengine high passed at 50Hz, and XO2 = tweeter reference from main speakers.
- Generate a Multiway WAV and logsweep both. Result as above.

I can see that the AudioEngine is 9010 samples (187.5ms) delayed with respect to the tweeters. I then time aligned both speakers.

View attachment 383447

Then I assigned all my DAC's to ASIO channels and enabled the virtual ASIO I/O channel.

View attachment 383448

Next step is to route all the channels appropriately. Look at that, I have 3 DAC's in the same ASIO channel! VB Matrix allows me to delay the AudioEngine, so I dialled in a 20ms delay.

View attachment 383449

Then I set up Acourate Convolver to send channels 1-8 to the main speakers, and channels 9-10 to the "ambience" speakers.

View attachment 383454

Then I had fun placing the "ambience" speakers in various positions in the room. My main speakers have a little gap at the back, perfect to hide a pair of small computer speakers, effectively converting it into an "almost" dipole. I am limited by cable length as to where I can place the speakers - the furthest I can place them apart is on either front corner of the room. Can't place them where I want just yet.

The effect is pretty remarkable - the 20ms delay was chosen because it lays outside the Haas fusion zone and early enough to be perceived as "ambience" but not so late that it is perceived as an "echo". It really does add spaciousness, and the effect is far more convincing than "concert hall" reverb.
I've experimented with this with a lot of conditions and methods in brir state (still experimenting), and my conclusion to date is to destroy the direct sound area as much as possible. It will be more helpful if you shoot at the wall or face the ceiling and play the speaker so it won't be recognized as normal hrtf. If you increase the physical distance more, the naturalness increases more. Of course, the properties of initial reflection are clearly present in space, so it's hard to overcome its limitations, but the stronger the sound absorption, more speakers, more angles, more height, more distance information you add, the more you won't feel the speaker, only the sound will remain.
 
I've experimented with this with a lot of conditions and methods in brir state (still experimenting), and my conclusion to date is to destroy the direct sound area as much as possible. It will be more helpful if you shoot at the wall or face the ceiling and play the speaker so it won't be recognized as normal hrtf. If you increase the physical distance more, the naturalness increases more. Of course, the properties of initial reflection are clearly present in space, so it's hard to overcome its limitations, but the stronger the sound absorption, more speakers, more angles, more height, more distance information you add, the more you won't feel the speaker, only the sound will remain.

What is brir state? Backward radiating ... IR?

I did some reading about direct vs. reverberant sound and the critical distance. What you seem to be saying is that you want the reverberant (ambient) sound to exceed the direct sound of the loudspeakers. To my knowledge I have not seen anybody advocate this. Surely you lose clarity if you were to do that?

Anyway, it's very early days. I have had to switch my system off (it's getting late) but I turned them down until they were "just noticeable" and the effect is quite pleasing. Tomorrow I will get my soldering iron out and join some spare lengths of speaker cable together and see if greater distance between the ambient speakers gives me better sound. So far this experiment will only cost me $20 ... which will be my donation to VB Audio.
 
What is brir state? Backward radiating ... IR?

I did some reading about direct vs. reverberant sound and the critical distance. What you seem to be saying is that you want the reverberant (ambient) sound to exceed the direct sound of the loudspeakers. To my knowledge I have not seen anybody advocate this. Surely you lose clarity if you were to do that?

Anyway, it's very early days. I have had to switch my system off (it's getting late) but I turned them down until they were "just noticeable" and the effect is quite pleasing. Tomorrow I will get my soldering iron out and join some spare lengths of speaker cable together and see if greater distance between the ambient speakers gives me better sound. So far this experiment will only cost me $20 ... which will be my donation to VB Audio.
Binaural impulse with room reaction (according to individual hrtf). Yes, no one in general wants the reflection to exceed the direct sound. But if you hear a sound in the center of a large space, you hear a reflection coming from far away, hitting a wall or ceiling. But it's for the effect of being dispersed. About 45 to 60 degrees, 120 to 135 degrees, even though the direct sound is clear (even if the speaker is facing me), it's less like there's a sound coming from that side. However, the direct sound being broken or attenuated was much more natural and more like a feeling of being filled with sound rather than a sound coming from that side.

This, of course, came to this conclusion when we tested it even in an environment where there are clear limitations depending on the size of the space and room acoustic for each distance, and where there are only regular rooms with reflections, direct sounds close to the anechoic room.
Your attempts are highly appreciated because fewer people add or copy reflexes like this.

++)
I wrote it down on my cell phone and turned on the computer.
I also just shared my experience by asking for advice from multiple users and testing it from different angles, heights, and distances.
I don't know which one is better.
But my goal was not to feel the speaker, but to feel the sound filled, or the any space or outdoor field itself.

What you seem to be saying is that you want the reverberant (ambient) sound to exceed the direct sound of the loudspeakers. To my knowledge I have not seen anybody advocate this. Surely you lose clarity if you were to do that?
Strictly speaking, I was talking about making the direct sound of the speaker for the reflection sound other than the main speaker transformed or heard less in some way so that it is less recognized.

1722178638294.png

1722178716836.png
 
Last edited:
@Keith_W, thank you for going into so much detail!

The effect is pretty remarkable - the 20ms delay was chosen because it lays outside the Haas fusion zone and early enough to be perceived as "ambience" but not so late that it is perceived as an "echo".

You probably already thought of this, but if the reflection path length from the rear-firing speakers to the listening position is longer than the direct sound's path to the listening position, any "surplus" path length adds to that 20 milliseconds of delay.

It really does add spaciousness, and the effect is far more convincing than "concert hall" reverb.

Just so I understand clearly, what do you mean by "'concert hall' reverb" in this context? Is that a setting on a processor?

... I turned them down until they were "just noticeable" and the effect is quite pleasing.

I have been doing something conceptually similar in some of my home audio designs, but without any added delay; in other words, with only the delay from the reflection path length for the rear-firing drivers. Ime the optimum loudness level for the rear-firing drivers does indeed seem to be right at the edge of where they start to affect clarity.
 
Last edited:
@Keith_W and @Duke, thank you for sharing very much interesting experiments and discussions!

You probably already thought of this, but if the reflection path length from the rear-firing speakers to the listening position is longer than the direct sound's path to the listening position, any "surplus" path length adds to that 20 milliseconds of delay.
Yes,
and furthermore, if we have sound reflection wall/plane near or behind our listening position we may have some automatic "reverberation" effects due to the possible standing waves (primary reflection, secondary reflection, and so on) in very much lower gain; I can/could objectively observe them using rather strong excitation of room air using rectangular tone-burst signals of any specific frequency, as I shared in my post #498 on my project thread.

Somewhat relating to the "objectives (i.e. intentionally added delay)" of present experiments by @Keith_W, I wrote there as follows;
I well know that our listening environment should not be quasi-anechoic, and we do need suitable preferable enjoyable sound reflections resonances and/or standing waves, but here I will not (should not) be going into the detailed discussion on "how much" and "how intensively" sound insulation/deadening should be applied in our listening room, an eternal and endless theme in home audio setup.

The on-going experiments by @Keith_W are really amazing and very much interesting, but I believe the given audible benefits and actual future implementation/tuning would be greatly dependent on individual room modes, i.e. room acoustic environments.


By the way, as for similar "intentional reflection/dispersion" in high frequency zone covered by tweeters and more importantly covered by narrow directivity supertweeters (intending better and improved sound-stage/sound-image), my recent experiments and actual implementation (in my specific setup in my room acoustic environments) would be of your reference and interest, I assume; this approach in high Fq zone would be relatively less dependent on room modes, I believe.
A new series of audio experiments on reflective wide-3D dispersion of super-tweeter sound using random-surface hard-heavy material:
Part-1_ Background, experimental settings, initial preliminary listening tests: #912

Part-2_ Comparison of catalogue specifications of metal horn super-tweeter (ST) FOSTEX T925A and YAMAHA Beryllium dome tweeter (TW) JA-0513; start of intensive listening sessions with wide-3D reflective dispersion of ST sound: #921

Part-3_ Listening evaluation of sound stage (sound image) using excellent-recording-quality lute duet tracks: #926

Part-3.1_ Listening evaluation of sound stage (sound image) using excellent-recording-quality jazz trio album: #927

Part-4_Provisional conclusion to use Case-2 reverse reflective dispersion setting in default daily music listening:
#929
 
Last edited:
You probably already thought of this, but if the reflection path length from the rear-firing speakers to the listening position is longer than the direct sound's path to the listening position, any "surplus" path length adds to that 20 milliseconds of delay.

Yeah, aware of that. But this was a rough as guts initial experiment. It's more of a proof of concept to show that it will work. I will make a more serious effort to tune them in. The three variables that are the most important for this effect to work are (1) volume, (2) direction, and (3) timing. None of these have been optimized.

The little AudioEngines are severely limited in volume. They were designed as nearfield computer speakers and they excel at that job. But at 3m away they really struggle. If I decide to make this permanent, I will need to get new speakers capable of more SPL.

Later today I will rummage through my cable bin to find some longer cable which I will solder together. That will give me more flexibility on where I can place my speakers. Once the speakers are placed, I will work on the timing.

I have another idea of making multiple echoes to make it more "reverb-like". I am thinking about how I am going to do that in Acourate - probably copy the finished XO several times, and giving each copy increasing delay and decreasing gain. Then merge all the copies of the XO into a single filter. I don't know if it is possible to do this in Acourate, but I know I can do it in my convolver.

Just so I understand clearly, what do you mean by "'concert hall' reverb" in this context? Is that a setting on a processor?

Yes. You will find it in AVR's, and you can also find it in JRiver.

I have been doing something conceptually similar in some of my home audio designs, but without any added delay; in other words, with only the delay from the reflection path length for the rear-firing drivers. Ime the optimum loudness level for the rear-firing drivers does indeed seem to be right at the edge of where they start to affect clarity.

I know :) We had a PM exchange some months ago where I described this same idea ... remember? :)
 
For some reason your link redirects to "2 Watt Zen amp for headphones".
Strange, looks same, try this one

 
Yeah, aware of that. But this was a rough as guts initial experiment. It's more of a proof of concept to show that it will work. I will make a more serious effort to tune them in. The three variables that are the most important for this effect to work are (1) volume, (2) direction, and (3) timing. None of these have been optimized.

The little AudioEngines are severely limited in volume. They were designed as nearfield computer speakers and they excel at that job. But at 3m away they really struggle. If I decide to make this permanent, I will need to get new speakers capable of more SPL.

Later today I will rummage through my cable bin to find some longer cable which I will solder together. That will give me more flexibility on where I can place my speakers. Once the speakers are placed, I will work on the timing.

I have another idea of making multiple echoes to make it more "reverb-like". I am thinking about how I am going to do that in Acourate - probably copy the finished XO several times, and giving each copy increasing delay and decreasing gain. Then merge all the copies of the XO into a single filter. I don't know if it is possible to do this in Acourate, but I know I can do it in my convolver.

Yes. You will find it in AVR's, and you can also find it in JRiver.

I know :) We had a PM exchange some months ago where I described this same idea ... remember? :)

At least three of Sonus Faber's more expensive speakers have included a user-adjustable rear-firing midwoof-and-tweeter module, but I think what you're doing may be an improvement in two areas (even before the further refinements you mention above): Your adjustable time delay, and your higher net direct-to-reflected sound ratio.

My recollection is that your main speakers are fairly directional, and imo this is desirable for this sort of multi-directional system. If one starts out with wide-pattern main speakers and then adds more reflection energy, ime the direct-to-reflected sound ratio can end up being a bit too low. But if the direct-to-reflected ratio is arguably on the high side to begin with, then adding more reflection energy is a step in the right direction in that area.
 
Last edited:
Back
Top Bottom