Keith_W system

[Keith_W]

7 hours ago, Andythiing said:

Keith I know your love of classical - how genre specific are your current settings? Did you optimise for one genre or is it pretty well perfect across genres? Just interested to know  how much changing of curves etc you plan on doing or do you see it as largely “set and forget”?

 

It's not genre specific, it's more recording specific. One filter sounds better for thin sounding recordings, the other sounds better for thick sounding recordings. About 10-20% of my collection are historic recordings from the 30's to the 50's and some of them are unlistenable on a modern system because the spurious high frequency spectra quickly causes fatigue. For those, I can turn on a denoiser VST to tame them. Amazing that even free VST's are more advanced and more effective than some of the more rudimentary consoles the pro's were using in the 80's and 90's. 

 

I could set and forget and both these settings sound good with all types of music. But I know that with a bit of fiddling I could get music to sound even better. 

[zydeco]

This thread should sit alongside some of the great threads outlining speaker & amplifier builds. Keith is going well beyond simple room equalisation to designing crossovers and then defining overall tonality - more or less a digital speaker designer. I have not had the good fortune to listen to the system but, having part of this digital EQ journey, it’s an amazing effort.

[Keith_W]

Ummm thank you but I would not go that far. If you read every post in this thread you can see that I have made a lot of mis-steps and wrong turns. In fact I think there may be an even older thread lurking around in SNA before I embarked on this digital journey where I made a lot of stupid and expensive mistakes.

 

You can see some remnants of that old journey in my current system, for example the Cary CAD-211AE amplifiers. You don't need 110W of push-pull amps to drive 98dB/W/m horns. I bought them because I thought I needed more amplifier power and still wanted to stay with valves. Back then, I thought that the specified 92dB/W/m was actually accurate and I would get adequate volume from a more powerful amplifier. Needless to say ... loudspeaker impedance, phase angles, current draw, etc. did not factor in my purchase decision because I didn't know anything about it. My thinking was - I need power, I want triodes, so I spent $20k on the most powerful triode amp you could buy for reasonable money. 

 

The journey to the current system has been stepwise, upgrading and buying new equipment every time an iteration did not work as well as I hoped. It went from a more powerful valve amp, to bi-amp (valve on top, SS on bottom), to adding subwoofers, to analog active crossover (with the passive still in place) ... it took me years! Finally we come to the start of this thread where we slowly get to a digital crossover. And then many more years (fortunately, without having to spend too much money now that I have the equipment) before I manage to refine it to what it is today. 

 

I am a bigger idiot and a slower learner than most. I have met some young guys (like @rand129678) who know more than I do and they are so bloody young. Not to mention Stuart of SGR who was REALLY young when I first met him (just barely out of his teens and going through uni!) but already knew then more than I know now. 

[rand129678]

29 minutes ago, zydeco said:

This thread should sit alongside some of the great threads outlining speaker & amplifier builds. Keith is going well beyond simple room equalisation to designing crossovers and then defining overall tonality - more or less a digital speaker designer. I have not had the good fortune to listen to the system but, having part of this digital EQ journey, it’s an amazing effort.

 

I first heard Keith's DSP active system many years ago and that experience is what sparked my own interest in DSP crossovers + room correction.

 

And at the time, what he was doing was so advanced (still is) and over my head - and it's taken me all these years just to be able to learn, to have a technical conversation with him or anyone about DSP, without looking like a complete idiot But the learning journey all started back on that first listen.

 

And the learning never stops !

 

He's being too humble. He's such a great asset to this hobby we love and the StereoNet Community.

 

A gentleman and a scholar.

 

[Keith_W]

By the way, I came up with a better method than Mitch for time alignment with Acourate. I will describe it here for other Acourate users (I think I am the only one on SNA though!). I don't know how other software packages like Dirac and Audiolense do it, perhaps you guys would like to chime in with a description of your method. It would help people who are considering Acourate vs. Dirac vs. Audiolense make a decision which way they should go. 

 

This is Mitch's method, as described in his book: 

 

1. Open up the crossovers for the left sub, right sub into curves 1, 2; left and right woofers into 3, 4; left and right mids into 5, 6. 

2. Rotate and save left and right mids by +1000 samples

3. Rotate and save left and right woofers by +2000 samples

4. Rotate and save left and right subs by +3000 samples 

5. Create a multichannel filter with all the rotated crossovers and do a sweep. 

6. Create a new working directory and copy the sweep into it. 

7. Create markers to help locate the individual channel peaks by loading the original (un-rotated crossovers) and applying rotation to it: -31768, -30768, and -29768 samples respectively. 

8. Zoom in and note the time discrepancy between the marker and the measurement. Do the same for all the drivers. 

9. Rotate all the crossovers by your measured delays and do a verification sweep. 

 

This is my method: 

 

1. Delay the tweeter by 1000 samples. 

2. Create a multichannel filter with only the tweeter delayed. 

3. To measure the sub, turn off the amp for the woofer and mid and do a sweep. (For me it is a simple matter of turning down the slider on my mixer). Do the same for all the other drivers. 

4. Acourate positions the tweeter at sample 6000. The pulse you are measuring should be at sample 5000 if it is time aligned. If you wish, you can load a tweeter crossover into another curve and rotate it by -1000 samples to act as a marker. If it isn't at position 5000, then note the discrepancy in terms of samples. If it is at 4990, it is 10 samples to the left (meaning too early); it needs to be delayed by 10 samples. 

5. Rotate all the crossovers by your measured delays and do a verification sweep. 

 

The advantage of this method is less faffing around with Acourate which is frankly a pain in the arse. 

 

If you suspect that your numbers are off, you can do this calculation: 

   - Conversion of samples into milliseconds = n/samplerate * 1000

   - I am using 48kHz sample rate, so a delay of 1 sample is 1/48000 * 1000 = 0.02083ms. 

   - Speed of sound is 343m/s. To convert milliseconds into distance in millimeters, it is 343 * T = 7.145mm. 

   - Measure the distance between your two drivers. If it is close to 7mm, you are in the right ballpark. Note that there are other factors that might contribute to the delay, in fact there are three: time of flight delay due to physical misalignment between the drivers, inertia, etc; electrical delay due to phase shifts at the crossover; and computing delay if you are implementing complex corrections. 

 

Aaaaaaanyway, getting back to my system. There are still a few things on my to-do list: 

 

1. Create a virtual double bass array. 

2. I have Ambiophonics installed as a VST. I need to play with it some more. 

 

The former is more faffing around with Acourate. The latter is much more difficult because you are not guided by measurements, you have to do it by ear. Oh yeah, and in case you haven't noticed ... all these upgrades are free. Well, free for me, since I have already paid for it  

Edited May 14, 2023 by Keith_W

[rand129678]

11 minutes ago, Keith_W said:

I don't know how other software packages like Dirac and Audiolense do it, perhaps you guys would like to chime in with a description of your method. It would help people who are considering Acourate vs. Dirac vs. Audiolense make a decision which way they should go. 

 

Mitch does a better job of explaining the process with Audiolense XO than I could ever do in writing (the XO version is the one that supports this driver time alignment and DSP crossover etc).

 

See his review of AL 5 (now at v6):

https://audiophilestyle.com/ca/ca-academy/audiolense-digital-loudspeaker-and-room-correction-software-walkthrough-r682/

 

A snippet below:

 

 

Edited May 14, 2023 by rand129678

[rand129678]

Not to pollute this thread but in case it helps someone to get started... that review by Mitch I linked above plus this one below , was enough to get me great results with Audiolense XO:

 

https://audiophilestyle.com/ca/ca-academy/ integrating-subwoofers-with-stereo-mains-using-audiolense-r712/

 

Then you chip away at improving slowly over time.

 

Every time I do a new measurement, I learn something and improve something

[Jventer]

Guys what you are talking about is way over my head, but I have a question.

For a noob like me is it not worth it to pay some $ and get Mitch to do the work for me?

[Keith_W]

Yeah, you could get either Mitch from Accurate Sound or Thierry from Home Audio Fidelity to do it for you. Both are likely to do a far better job than what I am able to do given their many years in the industry. But it all depends on what you want. For me, I get a sense of fulfilment from doing something difficult, learning something new, and doing it myself. I am happy to pay someone to do something that I can not do because it may require special skills (e.g. hire a lawyer, architect, buy art) or require special equipment (car mechanic, need anything made), or can't be bothered to do (paint the house, mow the lawn). But for everything else, I prefer to do it myself. There are good guides and online resources to help you along. 

 

One major reason for doing it yourself is that you can try different things and hear and measure the results for yourself as you learn more. I have done dozens of iterations of filters and I can load up old filters to remind myself of what the system sounded like a few years ago. I can tell you that once you learn of the possibilities of what you can build into a digital crossover, especially one with 130,000 taps, a powerful computer, and sophisticated software ... any passive crossover seems crude by comparison. Those crossovers only correct for the speaker and it corrects it in a very low-res way. A well designed digital crossover considers how the speaker interacts with the room, and it is NOT about simply applying EQ. Acourate can adjust the timing and phase of individual audio bands and get rid of nulls. 

 

I think the part that I struggled with the most was NOT acquiring the knowledge or using the software. What was most difficult was knowing whether it would even be feasible. Back when I started, there were very few people doing it, and it was harder to find good information on what to do. When Mitch came out with his eBook which contained a step-by-step guide to using Acourate, and his walkthrough on Audiophile Style, it was all the impetus that was needed. 

 

These days I am starting to think that Audiolense is a better package for beginners because it is more automated (see replies above). Even with this new version of Acourate, you end up having to do a lot of math to figure things out and it is a very manual process. I do not know which package offers you more functionality, because I have never used Audiolense, and the Audiolense users I do know have not gone as far as me (they are using passive XO's and doing overall speaker correction). I would love to know the answer. For ease of use, I would say that Audiolense wins hands down. For power users (or those who aspire to be power users) ... I honestly have no idea.

 

Uli Brueggemann (author of Acourate) is against automated processes because "it keeps users stupid" and because he thinks automated software might make mistakes (source), so I don't think Acourate will become easier to use in the future. His actual intention is for users to go through the process and have human eyes verify the measurements and make decisions. 

 

I have found that Acourate/Audiolense users are a passionate bunch and love helping each other. Even before I started, a guy named "dallasjustice" on another forum rang me from the USA to discuss how Acourate would work in my system. And this guy had no financial ties to Acourate. He was just extremely generous with his time. 

 

So ... if you want to embark on this journey and do it yourself, there are plenty of people willing to help and plenty of online resources. 

[AudioGeek]

I switchd from Accourate to Audiolense abut 18 minths ago.

If you dont want to read books, continuously troubleshoot etc - Audiolense all the way. If you want to dive in and spend hours on a computer then Accourate.

 

Have also used both Mitch's service and Home Audio Fidelity. Cant speak highly enough of both.

 

@Jventer Thierry's service is cheaper and all you need is a usb measurement mic and use his automated measuring tools. For Mitch you need to get audiolense, but he will walk you through it. Both can do mutichannel systems.

[Keith_W]

18 minutes ago, AudioGeek said:

I switchd from Accourate to Audiolense abut 18 minths ago.

If you dont want to read books, continuously troubleshoot etc - Audiolense all the way. If you want to dive in and spend hours on a computer then Accourate.

 

Yeah, I would agree with that. I have no issue with reading books, provided that I am learning about room acoustics, speaker and crossover design, and getting ideas on what I can improve. But I do not want to read books about how to use the software, and there is a bit too much of that required for Acourate. There are still big holes in my knowledge of Acourate, because there is no manual and the Acourate Wiki is painfully incomplete and it's written in "German English" - a rather terse writing style which assumes too much of the reader.

[rand129678]

I would say Audiolense XO is more automated than Acourate - but it is definitely not fully automated like the latest Dirac Live for example.

 

Acourate has so many features that would go unused by people even doing DSP crossover design and room correction.

 

So while Acourate has more features than Audiolense, in practical terms for most users, the more features might be irrelevant.

 

I was able to get this step response for a pair of 2-way speakers plus 2 subs, all 6 drivers time aligned.  This is both left and right speakers on one plot. The process would be identical even for 3-way speakers plus 4 subs.

 

And if I spend more time I can get the pre-ringing down further. Its not like Audiolense fully automates it for you and then you are stuck with that.

 

The 2 links I shared above of Mitch's Audiolense reviews shows you how you can tweak pretty much everything, to further optimise.

 

Having said all that, I still have an itch to scratch by trying Acourate, just for fun and giggles

 

 

Edited May 22, 2023 by rand129678

[frednork]

2 hours ago, rand129678 said:

I would say Audiolense XO is more automated than Acourate - but it is definitely not fully automated like the latest Dirac Live for example.

 

Acourate has so many features that would go unused by people even doing DSP crossover design and room correction.

 

So while Acourate has more features than Audiolense, in practical terms for most users, the more features might be irrelevant.

 

I was able to get this step response for a pair of 2-way speakers plus 2 subs, all 6 drivers time aligned.  This is both left and right speakers on one plot. The process would be identical even for 3-way speakers plus 4 subs.

 

And if I spend more time I can get the pre-ringing down further. Its not like Audiolense fully automates it for you and then you are stuck with that.

 

The 2 links I shared above of Mitch's Audiolense reviews shows you how you can tweak pretty much everything, to further optimise.

 

Having said all that, I still have an itch to scratch by trying Acourate, just for fun and giggles

 

Have you compared Dirac with AL?

[Keith_W]

The last time I looked (a few years ago), you have to use Dirac software to generate filters, and you can only run the filters on Dirac's convolver. Or any hardware that has Dirac's convolver built-in, like the MiniDSP and some Yamaha and Denon receivers. I don't like closed systems, it limits your choice. Simple fix for them, really ... build an export function into Dirac software, an import function into their convolver, and suddenly it's an open system. Have they done that yet? 

[rand129678]

This aspect of Dirac live is still the same.

 

However I did speak with a clever chap that managed to digitally record the impulse responses created by Dirac Live and was then able to generate convolver friendly files to use in HQPlayer (which takes in 64-bit mono WAV, like many other convolvers).

 

He did detail the process and it wasn't too hard but I was already 'all in' with Audiolense XO at that point.

 

There is hardware like the miniDSP you mentioned, which can accommodate both DSP crossover plus Dirac live for the room correction but these are typically limited to 4k tap (max I've seen is 8k) filters in total.

 

Mitch does a great explanation below of limitations of all hardware based DSP (minDSP, AVRs). Both Audiolense and Acourate can spit out 65k tap and 131k tap filters per channel

 

65k taps per channel at 48kHz is 48000/65000 = 0.7 Hz control per channel. Important for the lower frequency region - the most important region for room correction.

Edited May 15, 2023 by rand129678

[rand129678]

1 hour ago, frednork said:

Have you compared Dirac with AL?

 

I haven't but Mitch did a great review of the PC based Dirac Live 2 (older) and showed verification measurements, and he showed it's correction is up there with Acourate and Audiolense XO.

 

https://audiophilestyle.com/ca/reviews/dirac-live-2-digital-room-correction-software-walkthrough-r884/

 

The PC/Mac version of Dirac Live is believed to use ~32k tap filters, more than what you'll find in Dirac Live hardware solutions.

Edited May 15, 2023 by rand129678

[davewantsmoore]

10 hours ago, Jventer said:

For a noob like me is it not worth it to pay some $ and get Mitch to do the work for me?

Very much.

[frednork]

16 hours ago, rand129678 said:

 

I haven't but Mitch did a great review of the PC based Dirac Live 2 (older) and showed verification measurements, and he showed it's correction is up there with Acourate and Audiolense XO.

 

https://audiophilestyle.com/ca/reviews/dirac-live-2-digital-room-correction-software-walkthrough-r884/

 

The PC/Mac version of Dirac Live is believed to use ~32k tap filters, more than what you'll find in Dirac Live hardware solutions.

 

Yes, but he also said this here on ASR and elsewhere.

 

"Dirac uses a mix of IIR and FIR filters. IIR filters are used at the low frequencies and therefore offers no excess phase correction at low frequencies to correct for the rooms non-minimum phase response.

 

Don't get me wrong, I have reviewed Dirac extensively here. But because of these audible/measurable shortcomings, it did not make the SOTA list."

 

Things may have changed since then , I dont know.

 

 

 

 

Edited May 16, 2023 by frednork

[rand129678]

4 minutes ago, frednork said:

 

Yes, but he also said this here on ASR and elsewhere.

 

"Dirac uses a mix of IIR and FIR filters. IIR filters are used at the low frequencies and therefore offers no excess phase correction at low frequencies to correct for the rooms non-minimum phase response.

 

Don't get me wrong, I have reviewed Dirac extensively here. But because of these audible/measurable shortcomings, it did not make the SOTA list."

 

Things may have changed since then , I dont know.

 

 

 

 

 

Fair call. His post is 2021

Dirac Live Bass Control and Dirac Live Active Room Treatment are now here

 

Pretty advanced stuff from the info they've published.

 

I'm sure we'll get an updated review from Mitch.

 

[frednork]

1 minute ago, rand129678 said:

 

Fair call. His post is 2021

Dirac Live Bass Control and Dirac Live Active Room Treatment are now here

 

Pretty advanced stuff from the info they've published.

 

I'm sure we'll get an updated review from Mitch.

 

Yes indeed, have heard positive comments but a thorough review and comparison would be excellent.

[rand129678]

And to link it back to Keith's system or anyone else with DSP crossover active system

 

Since Dirac Live still doesn't offer DSP crossover support, 
would need to store the DSP crossover in a box, like miniDSP hardware offer.

 

Then you can run Dirac Live on PC

 

They have plugins too (VST2, VST3, AU, or AAX ) but would have to check if Bass Control and Active Treatment have come to the plugins yet

 

[Satanica]

25 minutes ago, rand129678 said:

They have plugins too (VST2, VST3, AU, or AAX ) but would have to check if Bass Control and Active Treatment have come to the plugins yet

 

Bass Control has but Active Room Treatment is still exclusive to Storm Audio processors (Beta).

[rand129678]

2 hours ago, frednork said:

Yes, but he also said this here on ASR and elsewhere.

 

"Dirac uses a mix of IIR and FIR filters. IIR filters are used at the low frequencies and therefore offers no excess phase correction at low frequencies to correct for the rooms non-minimum phase response.

 

Don't get me wrong, I have reviewed Dirac extensively here. But because of these audible/measurable shortcomings, it did not make the SOTA list."

 

Things may have changed since then , I dont know.

 

Looking at Dirac Live Bass Control website, it looks like this is directly addressing what Mitch mentions there in his 2021 post.

 

https://www.dirac.com/live/bass-control/

 

Compared to Acourate and Audiolense , Dirac is a bit of a black box, but Dirac have shown enough (via verification measurements) up to that point for me, showing they do what it says on the tin.

 

But for completeness, we'll wait for Mitch's followup in depth review with verification measurements.

 

I may do a trial of their desktop app  and compare with my Audiolense measurements. I would need something else like a miniDSP box to handle the DSP crossover though.

 

[Satanica]

30 minutes ago, rand129678 said:

I may do a trial of their desktop app  and compare with my Audiolense measurements. I would need something else like a miniDSP box to handle the DSP crossover though.

 

You can also use a professional USB interface and one of the advanced software players such as JRiver and Roon (I think) and presumably others.

Here's a professional USB interface with lots of outputs: https://focusrite.com/en/usb-audio-interface/scarlett/scarlett-18i20

[rand129678]

1 minute ago, Satanica said:

 

You can also use a professional USB interface and one of the advanced software players such as JRiver and Roon (I think) and presumably others.

Here's a professional USB interface with lots of outputs: https://focusrite.com/en/usb-audio-interface/scarlett/scarlett-18i20

 

I've had that Focusrite before and already have a multichannel DAC.

 

But the Focusrite can't store your DSP crossovers.

 

My comment was specific to using Dirac Live plugins and/or desktop app, since it doesn't support DSP crossover, you need to have your DSP crossover somewhere.

 

Inside a miniDSP box is a hardware solution , for the DSP crossover.

 

This would allow you to use Dirac Live with a DSP crossover active system.

 

As long as your software can support both convolution (for DSP crossover) as well as Dirac Live plugs, then that's another way to do it.

 

JRiver supports this but Roon doesn't support plugins.

 

Another method is using software to route audio between applications. Can get messy but definitely do-able.