ASR Acourate users

[CK.]

Yes, Acourate can import REW measurements. In REW, export the impulse response as a .WAV and open it in Acourate.

The problem with your setup is that you are using a MiniDSP with 2048 FIR taps. The heavy lifting with a MiniDSP is done with minphase IIR. Acourate is a specialist FIR tool. Whilst it CAN generate biquads for MiniDSP, its support is limited to a few models (I can't remember which). Because you already own an Acourate license, I suggest you do half the DSP programming in REW (PEQ generation) and the other half in Acourate (2048 FIR taps). Even then I suspect it would be a waste of time because 2048 taps is paltry, it's like drawing on a postage stamp with a crayon held between your toes.

But if you REALLY want to do it, your workflow would look something like this:

1. Make sure you read my eBook below about how to take measurements.
2. Apply any XO (if any) to your subs and main speakers via your MiniDSP.
3. Use REW to time and phase align your sub to your main speakers.
4. Use REW's EQ tool to generate PEQ's for your MiniDSP, then load them into the MiniDSP.

At this point you should have your PEQ's populated and you can then work on the FIR filters in Acourate.

5. Take a measurement of the subwoofer together with the left speaker, then a measurement of the right speaker alone (no sub). Our intention here is to use Acourate's ICPA function to shift phase bands in the bass to help remove any dips in the freq response.
6. In REW, export this measurement as mono .WAV then import them into Acourate. Change the sample rate to match your MiniDSP (for a MiniDSP Flex, it's 96kHz), then save this as Pulse96L/R.
7. Go through Acourate Macros 0-6 as normal (see my Acourate guide in the same Google drive folder as the REW book).
8. After the filters have been generated, open XO1L96/XO1R96 in Acourate, then use the "Cut N Window" command to decimate the taps down to 2048. You will observe that the shape will change to something really ugly and almost useless. When you see this, you will really question whether this exercise is worthwhile.
9. Save this as a .WAV. I don't know how to import FIR taps into a MiniDSP, you will have to consult your manual there.

If you want FIR, you are much better off getting a multichannel interface. You will need at least 4 channels.

Thank you for your detailed responce.

Let me explain a bit better what is my current situation, I am doing all the cross-over and PEQs using IIR in the miniDSP. I noticed that between 60-90hz there is a somewhat constant phase mismatch of about 160 degrees, and between 100-120hz about 60 degrees.

So I am trying to apply a very broad phase correction for these two bands with constant phase rotation. like an all-pass filter but with specific phase rotation.
Does this make sense? Is the 2048 taps not enough?

P.S I did read your acourate tutorial a while ago and it is very good, hopefully will bet a multichannel interface soon to explore things a bit more.

 

[ernestcarl]

Thanks for your prompt reply, basically any magnitude frequency correction is done using IIR. What I am looking to do, is basically change the phase of the sub at the XO poing, say rotate the phase between 60-90 Hz by 160 degrees and beween 100-120hz by 60 degrees. Because not they are almost complete;y out of phase. Is the 2048 taps still a problem in this situation?

Or you could adjust the phase of the mains themselves either with min phase xo and/or some phase EQ -- if you only have a miniDSP flex/2x4HD. Adjusting only the delays and polarity works well enough for most scenarios -- you can help visualize the issue to us by showing us your measurements -- overlayed sub+mains magnitude and phase -- with REW's alignment tool.

 

[CK.]

Or you could adjust the phase of the mains themselves either with min phase xo and/or some phase EQ -- if you only have a miniDSP flex/2x4HD. Adjusting only the delays and polarity works well enough for most scenarios -- you can help visualize the issue to us by showing us your measurements -- overlayed sub+mains magnitude and phase -- with REW's alignment tool.

The delays in the miniDSP have been adjusted so that the mains with sub are time aligned (based on the step responce) and the phase mismatch is not throughout. Sure I have attached the photos from REW if you could please take a look.
thanks

 

[ernestcarl]

The delays in the miniDSP have been adjusted so that the mains with sub are time aligned (based on the step responce) and the phase mismatch is not throughout. Sure I have attached the photos from REW if you could please take a look.
thanks

You could invert that narrow skew, but you might just get audible pre-ringing as a side-effect. If it were just a gentle nudge here and there, then the correction is usually transparent.

 

[3ll3d00d]

Thanks for your prompt reply, basically any magnitude frequency correction is done using IIR. What I am looking to do, is basically change the phase of the sub at the XO poing, say rotate the phase between 60-90 Hz by 160 degrees and beween 100-120hz by 60 degrees. Because not they are almost complete;y out of phase. Is the 2048 taps still a problem in this situation?

that sounds like you're asking for an all pass filter (which you can do with IIR)

 

[CK.]

You could invert that narrow skew, but you might just get audible pre-ringing as a side-effect. If it were just a gentle nudge hear and there, then the correction is usually transparent.

Yes I was thinking of using rephase for a couple of phase EQs (minding the pre-ringing), but in rephase I cannot properly design the filters while simulating the end result before applying. This is why I was askingi if accourate or any other (free) tool provides this capability.

 

[3ll3d00d]

if you really really really want to try using FIR manually then I'd suggest to use rephase and it's paragraphic eq to draw the response you want then attempt to produce a mixed phase version that gets as close to your designed shape as possible

alternatively perhaps brute force phase response manipulation is not what you want at all and it's some room setup issue that you need to address?

 

[3ll3d00d]

Yes I was thinking of using rephase for a couple of phase EQs (minding the pre-ringing), but in rephase I cannot properly design the filters while simulating the end result before applying. This is why I was askingi if accourate or any other (free) tool provides this capability.

it would be tedious and manual but you can certainly import the response from rew and the filter from rephase then convolve them to see what happens next

 

[CK.]

that sounds like you're asking for an all pass filter (which you can do with IIR)

Yes but an all pass filter has a fixed phase rotation, I was looking for variable FIR all-pass filter (rephase). but I was not able to simulate and see the result of a manual rephase filter so was looking for a more appropriate tool (this is why I referenced acourate)

 

[CK.]

if you really really really want to try using FIR manually then I'd suggest to use rephase and it's paragraphic eq to draw the response you want then attempt to produce a mixed phase version that gets as close to your designed shape as possible

alternatively perhaps brute force phase response manipulation is not what you want at all and it's some room setup issue that you need to address?

yes I was thinking along these lines of a mixed phase filter, but you make it sound like this is a bad idea Sure the phase distortion is due to the room I am pretty sure about that, but I cannot do anything about that I'm afraid (i.e. placement is fixed and bass traps are not an option).

I am not a rePhase expert, but if you can draw the target response and compare with my theoretical responce after applying the fitlers then yes that would be a way forward. just need to find out how you do this

 

[ernestcarl]

Yes but an all pass filter has a fixed phase rotation, I was looking for variable FIR all-pass filter (rephase). but I was not able to simulate and see the result of a manual rephase filter so was looking for a more appropriate tool (this is why I referenced acourate)

You could also try LinFIR if you use Windows.

*I know with rePhase, you have to go back and forth exporting and importing IRs with REW and manually convolving filters.

 

[CK.]

You could also try LinFIR if you use Windows.

hm Ok thanks let me look into this.

 

[ernestcarl]

hm Ok thanks let me look into this.

*note thought that's it's not really free if you want to export filters for actual use and not just simulation.

 

[3ll3d00d]

For what you are doing, it can be sufficient to eyeball it in rephase as you don't need ultimate precision to produce good enough alignment as these frequencies. Another approach is to dial in the iir filters while using an rta as you will see the phase response alignment via the magnitude response sum improving.

Another option is to try to use mso as the tool to generate both filters and the simulated response, again this is iir only so fits well for the dsp you have.

 

[CK.]

For what you are doing, it can be sufficient to eyeball it in rephase as you don't need ultimate precision to produce good enough alignment as these frequencies. Another approach is to dial in the iir filters while using an rta as you will see the phase response alignment via the magnitude response sum improving.

Another option is to try to use mso as the tool to generate both filters and the simulated response, again this is iir only so fits well for the dsp you have.

These are some good options I haven't thought about, cheers.

 

[OCA]

yes I was thinking along these lines of a mixed phase filter, but you make it sound like this is a bad idea Sure the phase distortion is due to the room I am pretty sure about that, but I cannot do anything about that I'm afraid (i.e. placement is fixed and bass traps are not an option).

I am not a rePhase expert, but if you can draw the target response and compare with my theoretical responce after applying the fitlers then yes that would be a way forward. just need to find out how you do this

You can use as many IIR allpass filters as you wish using biquads in miniDSP (can be designed in REW EQ) but they will all probably shift the phase in the opposite direction of what you really need. For time-inverted allpass filters, 2042 taps (not 2048) at 96kHz limits you to a min 47Hz frequency resolution and not much can be done in the region of interest even with rePhase optimization tricks.

 

[ernestcarl]

You can use as many IIR allpass filters as you wish using biquads in miniDSP (can be designed in REW EQ) but they will all probably shift the phase in the opposite direction of what you really need. For time-inverted allpass filters, 2042 taps (not 2048) at 96kHz limits you to a min 47Hz frequency resolution and not much can be done in the region of interest even with rePhase optimization tricks.

Ah, yes, 96kHz... eh, I keep forgetting that blasted fixed rate since I use 48kHz by default in JRiver. Would have been a little better if they had a 48kHz mode.

 

[john61ct]

You can use as many IIR allpass filters as you wish using biquads in miniDSP (can be designed in REW EQ) but they will all probably shift the phase in the opposite direction of what you really need.

What if active analog high SQ units handled the crossovers best as possible

and then use REW (with rePhase, MSO or other tools maybe Acourate) to create filters just handling the remaining issues at fewer taps so the limited processing of miniDSP can be more effective?

For context

Analog crossovers, only using DSP to treat issues the analog system cannot fix

it's all about balance There is nothing to balance. There is zero advantage to analog active filters, especially not if you already plan to use DSP anyway.

www.audiosciencereview.com

 

[john61ct]

Sorry for the dupe

The problem with your setup is that you are using a MiniDSP with 2048 FIR taps.

What if active analog high SQ units handled the crossovers best as possible

and then use REW (with rePhase, MSO or other tools maybe Acourate) to create filters just handling the remaining issues at fewer taps so the limited processing of miniDSP can be more effective?

For context

Analog crossovers, only using DSP to treat issues the analog system cannot fix

it's all about balance There is nothing to balance. There is zero advantage to analog active filters, especially not if you already plan to use DSP anyway.

www.audiosciencereview.com

 

[Keith_W]

What if active analog high SQ units handled the crossovers best as possible

And your answer was given by @voodooless in the other thread - it's a complete waste of time. A DSP crossover is superior to any active analog crossover, or any other type of crossover, period. The only advantage a PASSIVE analog crossover has over a DSP crossover is that only one stereo amplifier is needed. I get that for many people, having only one amplifier is an advantage. It's cheaper, easier to turn on, fewer boxes to hide, lower power bills. But if you have the budget for multiple amplifiers, a multichannel DAC, and some kind of computing device - it's DSP active XO all the way.

and then use REW (with rePhase, MSO or other tools maybe Acourate) to create filters just handling the remaining issues at fewer taps so the limited processing of miniDSP can be more effective?

If you are talking about the 2048 FIR taps on the MiniDSP - forget it. Please read this thread where I explain how FIR and IIR filters work. 2048 taps is too few to do anything meaningful. Opinions vary about how many taps you actually need, e.g. the author of LinFIR thinks that 16384 taps at 48kHz is enough (bin width = 48000/16384 = 2.9Hz) but Acourate typically uses 65536 taps at 48kHz (48000/65536 = 0.7Hz). MiniDSP typically has 2048 taps at 96kHz - which results in a bin width of 96000/2048 = 47Hz.