ASR Acourate users

[Keith_W]

Comparisons with DSPNexus?

DSPNexus is a fundamentally different DSP product. It's a hardware DSP unit. You need to design filters for it with filter design software. In the case of DSPNexus, they use Audio Weaver. DSPNexus uses 5th generation SHARC chips which use minimum-phase IIR filtering with a limited number of FIR taps. Competitors for the DSPNexus would include things like MiniDSP, Xilica processors, and so on.

Acourate is FIR filter design software which makes FIR filters. These can be linear phase or minimum phase, but most Acourate users make linear phase filters. The filters have to be loaded into a software convolver (this is a separate purchase, but if you use JRiver or Roon, you already have a convolver). This can be PC based, run on a headless SBC, or something like a Raspberry Pi. Competitors for Acourate are the other FIR filter design software packages, e.g. Audiolense, Focus Fidelity, Eclipse Audio's FIR Designer, etc.

A hardware DSP unit is better if you want to connect external sources for DSP because they usually have ADC's and digital inputs built-in. Whilst it is possible to run a phono stage through the ADC of your interface and route it through your software convolver, the process isn't as convenient or as seamless as it should be. On a MiniDSP (and maybe on a DSPN?), you press a button on a remote and you switch inputs. With software based DSP, you have to dive into your convolver's settings and change the signal routing. The real advantage of software based DSP is the virtually unlimited CPU power which lets you have thousands of FIR taps with very high sampling rates - up to 262,144 taps at 384kHz with Acourate.

Please read this thread: Understanding the state of the DSP Market for more info.

While this may be a bit off topic, one expert on DAC design at diyaudio.com, though unable to gain access to the DSPNexus's clocking, output stage, power supply schematics or at least looks at the pc boards, said that based on its top optional model AKM DAC chips the DSPN is a better value than the Merging Hapi for a multichannel DAC.

"Better value than a Merging Hapi" depends on your point of view. If all you need is 8 DAC channels, then of course the Merging Hapi is a very poor value proposition since it is so expensive. But if you need dozens of DAC channels, virtually unlimited expandability, a really robust system to tie all your devices together ... then it's hard to argue against Merging. Well, you could use Dante enabled devices which provide the same functionality, but they aren't much cheaper than Merging.

You don't have to use a Merging Hapi if you use Acourate though. Any multichannel DAC will do the job, as long as it has ASIO support. This includes things like Motu, RME, Lynx Hilo, Focusrite, and other pro audio interfaces.

In any case. my concern is that even if I hire Hollis Labs (as Danville's Al Clark advised) to remotely control my pc and guide me through making the live test tone measurements for them to build the software based crossovers, is the DSPNexus still the right solution to use for biamping my three-way speakers, either for semi noobs like me or expert users? That is, might the learning curve be substantially less with which other such 2 in/8 out box (es) or other processors?

All DIY DSP products have a learning curve, but some software is easier to learn than others. Audio Weaver does not look easy to learn, but it does look like an incredibly powerful tool with a lot of options that will confuse you (if you are a beginner) or delight you (if you are a DSP nerd). BTW I suggest you check their licensing, I think it's a subscription based software package with annual license required. Acourate is one off payment with lifetime license, but there are periodic upgrades, and if you want those upgrades you have to pay. Audiolense is the same.

 

[dped90]

It really is a toolbox, in that you have a bunch of various tools that you have to learn to use and deploy at the right time, a bit like deciding whether you need a screw or a nail to do the job. While both will nominally hold two pieces of wood together, they have advantages and disadvantages and it is up to you to decide which is better.

I managed to get through your first post at this thread and was impressed and even awed by the obvious power of this software. Unfortunately, I’m almost a totally ignorant DSP noob, who soon realized that daring to hope aboard the Acourate platform would quickly be drowning in numerous and largely incomprehensible filter design theories.

When someone at diyaudio.com suggested the DSPNexus, I had actually expected that learning Audio Weaver software would be easier than Acourate. But if it’s in fact harder beyond the first and second Audio Weaver tutorials I watched (see YouTube), then the DSPNexus interface is likely not an option, as Danville does not recommend use of it with any other software.

This is upsetting because while not intimately familiar with its clocking scheme, power supply and other design techniques, DAC expert https://www.diyaudio.com/community/members/markw4.373860/ claims the AKM AK4499EX DACs are currently the best available. And the DSPNexus has enough channels to support my four Rythmik F12 subwoofers.

Thus, assuming Davide is correct ( “…….an automatic procedure doesn't provide such a different result... for months of your life spared in studies and trials.”)-might Dirac ART or the software in fact work well with the DSPNexus multichannel DAC, for either staying with my main speakers’ passive crossovers or designing active ones to biamp them, plus room correction and multiple subwoofer management?

 

[3ll3d00d]

Seems completely off topic for this thread, I think comparisons of Dirac to other tools can be found in other threads if you search

 

[Carousel]

@dped90 Also on another thread... FYI... BACCH4Mac has an electronic crossover module in it. Only up to three-way and fourth order though. See: https://www.audiosciencereview.com/...cch4mac-short-review-first-impressions.39494/

 

[Roland68]

I managed to get through your first post at this thread and was impressed and even awed by the obvious power of this software. Unfortunately, I’m almost a totally ignorant DSP noob, who soon realized that daring to hope aboard the Acourate platform would quickly be drowning in numerous and largely incomprehensible filter design theories.

When someone at diyaudio.com suggested the DSPNexus, I had actually expected that learning Audio Weaver software would be easier than Acourate. But if it’s in fact harder beyond the first and second Audio Weaver tutorials I watched (see YouTube), then the DSPNexus interface is likely not an option, as Danville does not recommend use of it with any other software.

This is upsetting because while not intimately familiar with its clocking scheme, power supply and other design techniques, DAC expert https://www.diyaudio.com/community/members/markw4.373860/ claims the AKM AK4499EX DACs are currently the best available. And the DSPNexus has enough channels to support my four Rythmik F12 subwoofers.

Thus, assuming Davide is correct ( “…….an automatic procedure doesn't provide such a different result... for months of your life spared in studies and trials.”)-might Dirac ART or the software in fact work well with the DSPNexus multichannel DAC, for either staying with my main speakers’ passive crossovers or designing active ones to biamp them, plus room correction and multiple subwoofer management?

I'm a fan of the AK4499xx DAC chips and AKM myself, but I consider the best chip preference a preference, and I don't mean that in a technical sense. Whether ESS, AKM, Rohm, etc., the flagship DAC chips are all at a similar level, and it's much more about the implementation than the chip or manufacturer.

You'll have to familiarize yourself with the filters and familiarize yourself with each DSP, but that's a step-by-step process, and everyone grows with their requirements.
With Acourate, all you need is a PC with a sound card to get started, and you can use the trial version.
And that's often the biggest obstacle: just getting started.

 

[Keith_W]

I managed to get through your first post at this thread and was impressed and even awed by the obvious power of this software. Unfortunately, I’m almost a totally ignorant DSP noob, who soon realized that daring to hope aboard the Acourate platform would quickly be drowning in numerous and largely incomprehensible filter design theories.

Nah, it's pretty straightforward. I wrote a free Acourate guide and there's also @mitchco's book. My guide is more a collection of Acourate recipes with some explanation, and Mitch's book is a more thorough exploration of DSP, acoustics, and so on ... with a very easy to follow step by step Acourate guide. I am reasonably proficient with Acourate, and I can create a simple set of filters in about 30 minutes. But it would take me much longer than that to create a "refined" filter because of all the measurements involved.

The great thing about Acourate is how flexible it is. If you are starting off, you can create a set of filters with on-axis measurements and be done with it. But as you learn more, you can use virtually any measurement as the basis for your correction. One example: I am unable to take a 1m quasi-anechoic measurement of my woofer since the wavelengths are so long and the measurement is corrupted by reflections. So I take a nearfield measurement and simulate the baffle step. Then I convolve the baffle step simulation with the measurement and compare it to the actual 1m measurement. The result is a really clean measurement that I use as basis for my woofer correction. I described the method in more detail here.

IMO there is no other software which is as flexible as Acourate. All of them force you to work with the workflow pre-determined by the author. The exception is probably Audio Weaver. I can't comment on that in much detail since I don't use it, but it does look exceptionally powerful and flexible.

This is upsetting because while not intimately familiar with its clocking scheme, power supply and other design techniques, DAC expert https://www.diyaudio.com/community/members/markw4.373860/ claims the AKM AK4499EX DACs are currently the best available.

Don't worry about DAC's, just don't overspend on DAC's like I did. But then I didn't know any better at the time.

Thus, assuming Davide is correct ( “…….an automatic procedure doesn't provide such a different result... for months of your life spared in studies and trials.”)-might Dirac ART or the software in fact work well with the DSPNexus multichannel DAC, for either staying with my main speakers’ passive crossovers or designing active ones to biamp them, plus room correction and multiple subwoofer management?

Dirac is great if you don't want to learn the nuts and bolts of DSP, but then it isn't very flexible. If it makes the wrong decision, there's not much you can do about it. It's different with Acourate, because Acourate never makes the wrong decision because it doesn't make any decisions in the first place YOU make all the decisions, and if you don't like the result, you can go back and change it.

 

[dped90]

@dped90 Also on another thread... FYI... BACCH4Mac has an electronic crossover module in it. Only up to three-way and fourth order though. See: https://www.audiosciencereview.com/...cch4mac-short-review-first-impressions.39494/

Well, if I had a choice I'd rather stick with Windows hardware. But in any case, if your main speakers require two or more subwoofers like mine do, is there a software version of Bacch that helps you to design the active crossovers for the mains (or work with the existing passive crossovers) and also do subwoofer management, with any 8 channel DAC? Even this one? https://danvillesignal.com/dspnexus-dsp-audio-processor

 

[dped90]

With Acourate, all you need is a PC with a sound card to get started, and you can use the trial version.
And that's often the biggest obstacle: just getting started.

Indeed, that's a BIG advantage. I know this sounds silly but I wish I knew how capable I was to scale the Acourate learning curve high enough to at least design the best crossovers to biamp my two or three way pair of speakers. I guess I'll never know until I try, but I would likely need tons of instructions; months?

Also, I know that this probably sounds even sillier but my biggest fear of all with software based active crossovers is the risk of ear hair cell damage due to accidental high SPL exposure from my >97db/w/m speakers. I know that the Merging Hapi has auto mute protection to avoid this risk, and I think the DSPNexus does too. But what guarantee is there that after spending who knows how long learning enough to utilize Acourate for any basic purpose, I find that it's not sufficiently compatible with either of these two 8 channel DACs?

 

[Roland68]

Indeed, that's a BIG advantage. I know this sounds silly but I wish I knew how capable I was to scale the Acourate learning curve high enough to at least design the best crossovers to biamp my two or three way pair of speakers. I guess I'll never know until I try, but I would likely need tons of instructions; months?

Also, I know that this probably sounds even sillier but my biggest fear of all with software based active crossovers is the risk of ear hair cell damage due to accidental high SPL exposure from my >97db/w/m speakers. I know that the Merging Hapi has auto mute protection to avoid this risk, and I think the DSPNexus does too. But what guarantee is there that after spending who knows how long learning enough to utilize Acourate for any basic purpose, I find that it's not sufficiently compatible with either of these two 8 channel DACs?

For now, you only need a sound card with 6 or 8 outputs. There's plenty of experience with DACs, and since USB is the bridge, there shouldn't be any compatibility issues.

Two basic approaches:
You buy a pair of sacrificial speakers for 20-100 Euros to practice with, then reduce the volume using a voltage divider or volume control.
You simply leave the passive crossovers in the speakers for now, as that's the best protection, and only remove them once you're sure and have tested everything.

 

[Keith_W]

Indeed, that's a BIG advantage. I know this sounds silly but I wish I knew how capable I was to scale the Acourate learning curve high enough to at least design the best crossovers to biamp my two or three way pair of speakers. I guess I'll never know until I try, but I would likely need tons of instructions; months?

Read my Acourate guide which I linked to in my previous post. You don't have to create "ultimate" crossovers and filters on your first attempt, nor should you expect to. "Ultimate" only comes after a lot of refinement of your measurement technique - you need to be able to capture clean measurements and avoid correcting "bad" areas of your measurement - i.e. all the non-minimum-phase regions, or driver nonlinearities. It took me years to read and refine my technique, but it should take you a lot less given that I did it 10 years ago where there were much fewer online resources available. And also there are people like me around who will help you for free

I suggest the following iterations for a new user:

- Iteration 1: replicate your speaker's crossover in minimum-phase and linear-phase, and do room correction only. This will take you a few hours.
- Iteration 2: using the above XO, time align and do room correction. Another few hours.
- Iteration 3: linearise each driver with quasi-anechoic measurements using Acourate's driver linearisation macro. About an entire weekend.
- Iteration 4: explore the distortion profile of each driver, take directivity measurements, and refine the XO point and slopes.

After that you can start getting into the fun and crazy stuff to try to squeeze out the last few percent of performance. My system thread is full of such experiments.

Also, I know that this probably sounds even sillier but my biggest fear of all with software based active crossovers is the risk of ear hair cell damage due to accidental high SPL exposure from my >97db/w/m speakers. I know that the Merging Hapi has auto mute protection to avoid this risk, and I think the DSPNexus does too. But what guarantee is there that after spending who knows how long learning enough to utilize Acourate for any basic purpose, I find that it's not sufficiently compatible with either of these two 8 channel DACs?

The solution is fairly simple. All pro DAC's have mixers which limit volume output. Simply set that at a reasonable volume level. Even with a software glitch, your system will be incapable of exceeding a certain SPL. Another failsafe is to have volume control for each amplifier in your signal chain.

 

[3ll3d00d]

I think you need to be clear about your goals

You mention 4 subs so i took that to mean you need a mechanism to combine those in a way that gives you a good result from the combined output of the subs.

If so, acourate can host such a filter, eg created using mso, but it cannot design one.

 

[dped90]

@dped90 Also on another thread... FYI... BACCH4Mac has an electronic crossover module in it. Only up to three-way and fourth order though. See: https://www.audiosciencereview.com/...cch4mac-short-review-first-impressions.39494/

Small wonder why Audio Weaver is so powerful.

AudioWeaver — BACCH Labs

www.bacch.com

 

[landco]

The great thing about Acourate is how flexible it is. If you are starting off, you can create a set of filters with on-axis measurements and be done with it. But as you learn more, you can use virtually any measurement as the basis for your correction.

There is a free Room EQ Wizard program, which also allows you to create filters with a sufficient level of knowledge. Why should I buy Acourate?

REW is a cross-platform application, while Acourate is only for Windows. That is, the author asks for $ 450 for his program, but for all these years he was too lazy to make a version for Mac OS and Linux.

Maybe making a quality product to the author was too lazy? And he begins to tell you that the “years” are needed to create a good filter

 

[Keith_W]

There is a free Room EQ Wizard program, which also allows you to create filters with a sufficient level of knowledge. Why should I buy Acourate?

Sure, go ahead and use REW to make linear phase filters for a multi-channel system. It's not as if it can't be done. It's possible. After you finish, report back and let us know how it went. In particular, I would be really interested to see how you go with pre-ringing compensation and excess phase compensation. I have tried to figure out how to do it with REW with rePhase, and I failed. I have read every resource on rePhase that is available to try to get to the bottom of it. So if you succeed, I would love to know how you do it.

One major weakness with REW is that it does not have a built-in multichannel convolver. So, let's say you make crossover filters for a 3 way speaker and you want to test it. You have to get REW to save 6 .WAV files with the correct filter length, sampling rate, and impulse position. Find yourself a third party convolver, write a config file for it, then load the .WAV files. Route REW's output through the convolver to take a measurement. If you want to do this with ASIO, you will need to set up some kind of loopback. Doing all this would take me a couple of minutes with Acourate, but it would take more than an hour with REW. It's not impossible, but it isn't easy. I know, because I have tried.

The same can't be said if we are talking about minimum-phase IIR biquads for room correction. REW is much better at doing this than Acourate. Biquads need to be specifically tailored for DSP hardware, because biquads have poor portability. This is why REW's EQ designer has a drop down box with different manufacturers listed. So, if you are using MiniDSP, REW should be the tool of choice. In fact I would choose REW over Dirac, and I would not use Acourate. In fact, I recently helped a friend dial in his MiniDSP. I could have used Dirac (he has a Dirac license), Acourate, or REW. I chose REW.

 

[landco]

The same can't be said if we are talking about minimum-phase IIR biquads for room correction

Excuse me, but by default it is only about the correction of the acoustics of the room. To create active crossovers for speakers, there are other means, much more convenient tools. For example, the DBX Driverack PA2 system

 

[Keith_W]

Excuse me, but by default it is only about the correction of the acoustics of the room. To create active crossovers for speakers, there are other means, much more convenient tools. For example, the DBX Driverack PA2 system

Sure. Go ahead and use that. It's not linear phase though. I have no need to engage with you any more given your attitude.

 

[landco]

It's not linear phase though

Why do you replicate the audio-filled myths? A person does not hear phase distortions, so there is no point in chasing this ghost.

 

[Keith_W]

1. It is incorrect that we do not hear phase distortion. We can hear phase deviations if they are >30deg per ERB. See this post by JJ.

2. Linear phase is much easier to design since you can manipulate magnitude independent of phase. For example, LPF's and HPF's maintain the linear phase characteristic throughout, which means there are no summation problems. Contrast this with minphase, where you have to invert the polarity to get some LPF/HPF combos to sum properly (e.g. Butterworth 4th order), and even then it sums with a 3dB peak. With linear phase, the only consideration is the XO point and slope. No need to worry about the 90deg phase rotation per order with a minphase XO.

3. You might say that even if it electrically sums to flat, the minphase characteristic of drivers means they may not acoustically sum to flat. Well, with linear phase, you can linearise the amplitude and phase of the driver across the entire bandpass, which means they will acoustically sum to flat.

4. Bass correction is much simpler with linear phase. With minphase biquads, each PEQ you add also rotates phase, which is why you need MSO and DLBC to help you compute which biquads to use.

5. It is possible to perform excess phase correction with linear phase by time reversed all-pass filters. Because IIR biquads are causal, this is not possible. Well, maybe I should be careful about saying that because there are newer methods with delays and subtractive filters that can bestow some linear phase characteristics with IIR filters, but you will have to manually calculate the coefficients.

6. Because of the recursive nature of biquads, cascading biquads may lead to filter instability.

7. Because of quantisation errors (meaning, biquads may not completely describe a phenomenon), cascading biquads also increases noise.

 

[landco]

1. It is incorrect that we do not hear phase distortion. We can hear phase deviations if they are >30deg per ERB. See this post by JJ.

2. Linear phase is much easier to design since you can manipulate magnitude independent of phase. For example, LPF's and HPF's maintain the linear phase characteristic throughout, which means there are no summation problems. Contrast this with minphase, where you have to invert the polarity to get some LPF/HPF combos to sum properly (e.g. Butterworth 4th order), and even then it sums with a 3dB peak. With linear phase, the only consideration is the XO point and slope. No need to worry about the 90deg phase rotation per order with a minphase XO.

3. You might say that even if it electrically sums to flat, the minphase characteristic of drivers means they may not acoustically sum to flat. Well, with linear phase, you can linearise the amplitude and phase of the driver across the entire bandpass, which means they will acoustically sum to flat.

4. Bass correction is much simpler with linear phase. With minphase biquads, each PEQ you add also rotates phase, which is why you need MSO and DLBC to help you compute which biquads to use.

5. It is possible to perform excess phase correction with linear phase by time reversed all-pass filters. Because IIR biquads are causal, this is not possible. Well, maybe I should be careful about saying that because there are newer methods with delays and subtractive filters that can bestow some linear phase characteristics with IIR filters, but you will have to manually calculate the coefficients.

6. Because of the recursive nature of biquads, cascading biquads may lead to filter instability.

7. Because of quantisation errors (meaning, biquads may not completely describe a phenomenon), cascading biquads also increases noise.

Your theory has nothing to do with practice, since comparative blind tests do not confirm this.

 

[Keith_W]

Your theory has nothing to do with practice, since comparative blind tests do not confirm this.

Not my theory. You can go argue with JJ. I'm done with you.