The making of: The Two Towers (a 25 driver Full Range line array)

I've got to quote myself here, pretty pleased with seeing the results of the APL TDA measuring software on my home brew correction. Mr. Raimonds Skuruls encouraged me to try a demo of his high resolution measuring software. It would show more resolution than REW's wavelet view.
So I decided to take up that challenge (as it felt as a challenge for me, I did try and find my own way to audio nirvana, his software could prove me wrong)
So I decided to take the plunge. I was a bit scared at first. What if this software would prove my DSP processing to be wrong. Anyway, I decided I could learn a thing or two just by trying it.

I started as usual, outlining my microphone in REW, making sure my mic is in the sweetspot.
The resulting Wavelet graph in REW:
REW%20vs%20APL.jpg


Next I moved on to the demo version of APL. Installed the required software, including the Matlab compiler and ran a few tests.
At fist I did get odd results at 96000. So I decided to try 48000. My difficulty was getting APL to run trough JRiver's processing. Not the fault of the APL software, just needed to find out how to get it to work. If you look at the uncorrected graph, I was seeing a lot of that. Until I found the right key to get it to run trough JRiver's correction.

Once I got there, confirming the audio route visually in JRiver I got these screen grabs:
APL_Demo_wesayso.jpg


And in 2D view:
APL_Demo_Wesayso2D.jpg


Mr. Raimonds Skuruls, would you say this is a reasonable result, now viewed in higher resolution? :)

I did already say I recognised much of my own priorities in your paper. A way of looking at this audio problem, actually. I enjoyed reading the discussion you had with Toole. I recognised a lot of it in my own rationale, where I differed from Toole's opinion. At least what he did make public in his paper.
I just used different software to get this, the stuff I had available to get me there. To me it does look like it works either way. As it should!

Even the listening impressions you quoted from a client(*) did sound very familiar ;). So I figured I should be alright running the Demo. Now you can see the minimum phase behaviour I spoke of at the low end. More pleasing than linear phase down to DC.

To show what my result is without my FIR correction:
APL_Demo_wesayso%20no%20cor.jpg

I keep claiming on this forum this stuff really works, I hope more people do try it.

(*) = on the Prosound website

Fun Demo! Thanks for the opportunity. Basically it seems we agree on a lot of grounds. At least I like to think that. It took me the bigger part of this your to get this. Using a lot of my time to find what works and what doesn't. Spitting trough all graphs in REW to find my answers and using DRC-FIR outside of it's intended scope. The better the measurements became, the more pleasing sound I got. With all genre's I listened to.
No brute force correction that only works at a single position. I'm using short frequency dependant windows (as short as I could get away with) to do the correction.

Disclaimer: All measurements were made in a "live" living room at the listening position. What I show here is a Stereo measurement of my corrected line arrays in my room. Corrected with REW, DRC-FIR and JRiver and a lot of work!
Source of post: http://www.diyaudio.com/forums/full-range/284916-room-correction-speaker-correction-what-can-we-do-dsp-power-now-availabl-2.html#post4579120
 
Last edited:
Wesayso,
One question about your line array design. What made you choose the straight vertical line array over a curved array, was that purely a space consideration or did you have something else in mind. Perhaps a curved array would have just been a pita as far as building the enclosure, just curious.

Kindhornman,

My choice for straight arrays was a deliberate choice. By now we've all seen the CBT plots from Keele, and arrays bent the other way have come and gone too on this forum.
My reason was not related to the building of the enclosure. See the first couple of pages of this thread to realise that wouldn't have stopped me.
My reasoning has to do with line behaviour and the room vs the listening position. A full length array in a living room could be seen as an infinite long array if we apply Keele's floor mirror and add the appropriate ceiling mirror as well. Those mirror images are missing in action in Keele's paper but in my case I wouldn't know good a reason why it should not apply here. That would make his predicted plots for straight arrays look quite different. He did remember to put in the floor mirror for the plots regarding CBT array.

You could even see the "floor and ceiling mirror images" from a straight floor to ceiling array as SPL level shaded by default (perhaps even time delayed).

But the "seemingly" infinite array has other advantages. You get the -3 dB per doubling distance over a very wide frequency range. Meaning it actually has line array behaviour over that wide area. I'm sure we'll see something about that soon on a different thread.
There are also no efficiency penalties because no shading of the drivers in the array is done. So each driver only does it's own little share. Together they have a huge potential.

Wherever you are in the room, the sound will be quite equal, no crossovers to mess with timing to create lobes etc. See the papers from Dave Smith on his research about floor to ceiling arrays. Very agreeable information in that research.
In a room, looking at reflections it makes a lot of sense not to shade such an array. Move up or down, left to right and you'll still hear the same sound balance. Only a hand full of speakers that do that in a room. Get off axis on a 2 way or 3 way and your relative distances to the drivers playing different parts of the frequency spectrum messes up the relative timing.

Look at what the array sends out in the room, the power response. Even that will be quite similar to the on axis response. In all directions! (well except floor and ceiling, eliminating those is a huge advantage)
This is just a hand full of loose thoughts as of why I chose straight arrays and figured why they should work going in to this project. Yes, comb filtering does occur. But I haven't heard it yet. In all honesty you can find it in measurements if you're looking for it.
The additional FIR filtering does a proper job to get pretty spectacular results.
The arrays are a different kind of animal without that FIR correction. I wanted to find the limits of what was possible and most important: sound natural.

Fir correction does it's job well both in sound as well as graphs. All the plots presented by me are taken in my living room, a pretty normal room with only 3 damping panels with only one of them visible as a poster. Still the plots show excellent results.
To me that means that the compromises that line arrays bring are well worth the advantages they have over other type of speakers. Couple that to a small foot print, placement close to the walls and spectacular imaging and dynamics. I'm pretty sure I made the right choices. :D
It isn't as easy as run some software and boom! Great sound. It took a lot more effort than that. My goal was maximum results from minimal correction. Overcorrection, be it IIR EQ or FIR never sounds really good. The frequency dependant windows used do a good job of protecting you from trying to boost nulls and stuff like that. The sound is natural, an aural hologram if you will of what is in the recording.

I don't know if you read Patrick Bateman's review about the B&O speakers.
Small recordings sound small, big recordings present an equally big image etc. You get all this from 2 speakers that fit on an A4 sheet of paper each.

No regrets here. Better results than I could dream of.
 
wesayso, any chance you could do a short summary of what you did?
How did you measure (multiple points, single point, in-room, outside)?
How did you process the measurements in REW?

Thanks!

The short story of EQ hmm? :D

All measurements were taken inside my living room. FIR filtering based on a single point measurement. Verified result over a large listening area by measurements.

Measurements taken in REW with JRiver in the measurement chain at all times.

In JRiver PEQ's are set to get the rough raw response of the arrays within a plus/minus 7 dB window. Smooth broad EQ strokes, highest Q=3. Plenty of boost though at both extreme ends. After all, 25 drivers share that load so each one of them only has to do a tiny part.

The REW measured IR is exported to 32 bit mono wave file (one for each channel) and converted to RAW PCM. That PCM is the base for DRC-FIR with my own template and settings, created over time, optimised to suit my speakers and room. Same correction settings for left and right array.

Damping panels at first reflection points as much as I was allowed to by my significant other. Essential to get good results.

The special brew of settings, outside of the normal scope of DRC-FIR got me where I am today. Lot's of other experiments done as well, such as conjugation networks, mid-side processing, different wires, different sound sources, cross talk cancellation. Ambient back channels. But to learn more, just read this thread. :D

Imagine how long it will take to read it all. Then imagine how long it took me to get it all done! It was well worth it for me, is it worth it to you? ;)

It's practically full disclosure in this thread. I've read more than 10000x of what I've written, pretty sure of that! :rolleyes:
You name it, I've studied it. Toole, Geddes, Danley, Dunlavy, Smith, Jones, Linkwitz, the smart folks on here and many more, I've tried to get some out of all of them. I put it in a blender and come up with my own view with a bit of myself mixed in of coarse. I'm stubborn enough to have an opinion too :D.

Judging the graph below, it looks like it was well worth it:
TDA_3D.jpg

It sounds good too, after all, that's what counts!
 
Last edited:

pnix

Member
2014-10-16 5:09 pm
Thanks wesayso.

I did read most of this thread in the past. Just wanted to know if you had done something new or special with regards to room correction.

So basically what you did is a minimum phase correction with low Q filters based on a single point steady-state response with the resulting filter being the same for L and R. Correct? I'm just not clear what kind of FIR filtering you've applied as the last step. Magnitude? Phase? Both?
 
Last edited:
Definitely a combination of both. There is some phase correction involved.
The pré EQ was only applied to get the raw output of line arrays within the working limits of the DRC-FIR correction scope.
It works with the large number of drivers I use, I wouldn't recommend it on single driver setups. Line arrays need EQ to correct the output. I worked on finding the best balance of that EQ, treating my speakers and room as one system. Brute force correction will not sound that great. At least not outside of the sweet spot. I kept an eye on what the correction does outside of that sweet spot. After all, I'm not alone in this room all the time. :p
So the key was to find settings that work globally.
 

fluid

Member
2009-01-24 2:20 pm
Hi wesayso,

Really enjoyed reading through your thread and seeing the progress from build to FIR correction. Took me a long time to read it all but was well worth it. Before reading this I was somewhat against the idea of room correction but your use of the room correction tools to correct more of the speaker and less of the room seems to have great potential as shown in your measurements.

Have you considered trying to take some power response measurements?
I think these would be very useful and would help to show the directivity of the arrays in room. Much has been talked about the theory of line arrays for avoiding floor and ceiling reflections and maintaining wide horizontal directivity and I think a power response measurement would show this well.

I saw a video from Earl Geddes where he showed the power response vs the on axis response for various speakers as part of a presentation and the information was quite interesting.

Your speakers seem to measure well in most plots so a power response would be very interesting;)
 
Fluid, I'm glad you enjoyed the read.

It would be good to see the polar response of arrays like these, I agree with that. But you've got to picture these huge arrays, and a lot of stuff in my room as it is still a normally filled living room. I can't easily measure polars as a consequence. So I measure what I can.
The measurements along the listening area are what I'm restricting myself to.
I use the gain from wall boundaries at low frequencies. So basically I need them. The left and right speaker show differences as the setup isn't symmetrical.
I think for an honest view we're probably better off getting these type of measurements from the corner arrays build by ra7.
Power response is of interest to see how the direct response (as far as we can speak of direct response measuring it in a room) differs from the response "added" by the room reflections etc. I have shown examples of that with my multiple measurements along the listening axis.
Not completely accurate of coarse but a rough indication. It was even from before 2 installing 2 extra damping panels.
What I did was gate the average of those listening area measurements from 0 to 20 ms (and pretend it is the direct response I'm seeing there) and compare it to the average plot from 20 ms start to 500 ms.
soundpower2.jpg

Taken a long time ago with entirely different settings and as sais, without 2 extra damping panels that I installed later on.

I should probably revisit this at some point. But I'm also cutting loose from this project (declare it finished) to focus on more important stuff.

As said at the beginning of this thread. I had spare time to do all this crazy stuff because I found myself without a job after working over 24 years.
Though I did try to get people to hire me, in other words; find me a new job, it hasn't happened yet.
2 mayor reasons for that as I see it.
1) I'm not the youngest anymore, and there are a lot of people looking for jobs that have more paperwork than I do to prove their abilities.
2) I'm cursed with a chronic disease, meaning I can't work full time. I just do not have that kind of energy.

Number 2 has actually been the main reason I didn't find a new job. Without that I could have had quite a few new jobs already. Seems like nobody wants to hire you when you're a tiny bit "broken".

Sad as it may be, it's a reality for me. Currently I have no income. So I need to find something for myself to change that.

I've long thought about turning this hobby into a profitable income. Not looking to get rich, just enough to get by. I'm still hoping that one day that might come true for me. That would really be a dream come true. This hobby is the one thing where I get to do all that I'd like to do,
Analyse and solve problems (basically thinking :)), work with my hands, use my computer skills etc.

But right now I need to be realistic and make money where I can. Something like IT consulting as that is closest to what I did before in my paid job. Even though I'm addicted to this stuff, and can think of a million more tests I'd like to do, I have to draw a line at some point. I need to focus on starting my own business and make some real money. It's no fun to live on fumes, spending your life-savings. That doesn't work for very long.

So either I'd need sponsors or a paying boss to continue these tests or I need to quit and start making money doing something different from this all.

It has been fun, and I'm not going anywhere. But I can't continue spending my days doing fun tests, however much I love doing them. it's just not paying the bills!

So anyone up for a pair of line arrays, hand build? I can make you a pair, (lol) wrong market advertising on a DIY forum :D.

(probably not the answer you were looking for, sorry for that ;))

Edit: I just can't see there being a huge (local) market for hand build line arrays, costing about € 22.000,- just to cover costs and the client having to wait a year for me to build a pair. (Or buy my current arrays for that price and I'll build new ones) For a while I thought that might work, but I'm basically a nobody, no name to flaunt to sell something like this. The people in a market like that usually care more about the layers of lacquer and the "big name prestige" than the actual interest in it's performance.
 
Last edited:

fluid

Member
2009-01-24 2:20 pm
wesayso totally understand your position and am sorry to hear that you are still out of work. I had a fairly long break from work myself recently when I finished off a lot of my audio projects that I had collected over the previous few years when I had no time to make them but money to buy them :rolleyes:

I can understand why you want to draw a line under the project as there doesn't seem to be much more that you could do to improve what you have, different maybe but probably not better. That was the reason why I thought yours might be a good candidate for power response too:)

The graph you show there has some pretty good correlation so maybe we can infer that the power response if pretty good:D

John Atkinson from stereophile uses a fairly simple method to measure what he calls in room amplitude response
Measuring Loudspeakers, Part Three Page 8 | Stereophile.com

This is the method I would use if I was to give it a try as it could be done without having to move anything around ;)

Good luck with the job hunting, maybe you can show your project as an example of how you can solve problems and don't give up until you get the result you want:D
 
Definitely on my agenda, perceval. I'll look into all things that could earn me some money.

Who knows, I might even start selling 3D printable tweeter pods like these (for Car Audio):
tweeterpod-Hanatsu-cl.jpg

The Viva XT25 SC 90 fits in there perfectly and it improves it's dispersion.

Based on my own 3D printed pods:
xt25.jpg

Which could also turn into a sellable item.

Now back to the power response. I know it to be quite good, based on what I measured so far in my 25+ measurements and their results. If I have the mic out any time soon I'll see what I can do to prove it.

I have a demo of my system coming up soon that will probably give me a review to post (I hope). I'll post it, good or bad (I'm confident in the outcome).
It will tell you guys more than my own subjective view. I'm biased.
 

pnix

Member
2014-10-16 5:09 pm
Definitely a combination of both. There is some phase correction involved.
The pré EQ was only applied to get the raw output of line arrays within the working limits of the DRC-FIR correction scope.
It works with the large number of drivers I use, I wouldn't recommend it on single driver setups. Line arrays need EQ to correct the output. I worked on finding the best balance of that EQ, treating my speakers and room as one system. Brute force correction will not sound that great. At least not outside of the sweet spot. I kept an eye on what the correction does outside of that sweet spot. After all, I'm not alone in this room all the time. :p
So the key was to find settings that work globally.

Can you elaborate on the settings you used for the FIR correction? I'm familiar with tools like Acourate but not with Denis Sbragion's DRC which is what I assume you're using?
 
Can you elaborate on the settings you used for the FIR correction? I'm familiar with tools like Acourate but not with Denis Sbragion's DRC which is what I assume you're using?

pnix,

that would be one heck of a post since the information is scattered all over this thread... still you can have a look at previous pages of posts, and most talk about DRC is within a few adjacent pages. A quick search within the thread would come up with lots of info.

There's also another thread started by gmad that talks about convolution, DRC and settings. That might get you going as well. Here:

http://www.diyaudio.com/forums/full...ectrical-loudspeaker-correction-networks.html

wesayso has given so much of himself in this thread, but he now needs to look at other things to get going in his life.
 
I do indeed use Denis Sbragion's DRC. I've covered the (important) settings both in this thread as well as gmad's http://www.diyaudio.com/forums/full-range/275730-convolution-based-alternative-electrical-loudspeaker-correction-networks.html.

Just use the "search this thread" tool, go "advanced" and look for posts made by me and some clever search words and it shouldn't take long to find what you're looking for.

But, and that's a big but... the settings I use work well with my speakers in my room. Because that's what they were optimised for. The general ideas may apply to other kind of speakers as well. But I have doubts there's a one size fits all solution. Remember the particulars of line arrays. They do not compare well to other speakers in behaviour. They are a strange breed, that's why I like them!
I did change a couple of things in settings over the last few months and would not exclude minor changes in the near future. Work trough this thread plus the one I linked and you get a clearer view on the "why". I believe that part is much more valuable than knowing my actual settings which probably only work properly in my room, with my room treatment and speakers.

Perceval - You beat me to it, we must have posted at nearly the same time and gave nearly the same answer :D
 
Last edited: