Digital crossovers / room correction

[morricab]

Hi Brad,

Which dacs are you using with the mini dsp?

Hi Bill
I have a Monarchy M24 DAC on the highs and a Metrum Octave DAC on the lows.

 

[Zero000]

Hi Justin,

Sorry I was referring to the precise way you used it - did you correct the entire frequency response, which filters did you use, which convolver software?

I am well versed on Acourate so know the field quite well.

Literally as I said. It was a free PC demo version, and early one, s/w only. I used it to explore the effect of RC and FR corrections made by the software. I only played with the FR slopes a bit by dragging the plotted curves.

Here's what I feel about the room correction thing.

When I play my acoustic guitar, it sounds different in every room I play it. The interaction of the acoustic with the room is an entirely natural thing.

When studio music is recorded, a lot of care is used to eliminate room effects and many instruments etc are plugged directly into some studio hardware.

So doesn't it follow that I don't really need RC since room interaction is an entirely natural process for real instruments? I think so. And I hear so.

DIRAC is good at removing room artifacts but in the process strips the music of life and vitality. Such that it just sounds like it is coming from the speakers. It's clever, I don't deny it, but another issue is this. The maths processing used to do it is NOT transparent.

What I found when I was re-doing my Apogee crossover was this. I would use Peace/Equaliser APO to implement the notch I wanted to implement in the digital domain. I would model what I wanted by taking FR measurements and using XSim to work out what the circuit needed to be to effect the change. Those apps are just PC apps - Google them they are free and excellent.

Here's the crunch - when I prototyped the circuit in hardware using cheap components to start with, the hardware implementation sounded a lot better than the digitally applied PEQ.

The more PEQ you apply, the less transparent it becomes. That's just FR correction, though. Add RC processing in there too and that's a lot of mathematical mangling going on which has adverse effects on SQ. It's just the way it is with digital filters, I am afraid, though some implementations will be better than others.

 

[Zero000]

Bill IIRC I think there where some filter options in a combo box somewhere but I really can't remember what they were.

But since I have basically rejected RC on the above basis i.e. the speakers are the musical instrument and should interact with the room, I'm not too worried about it.

If you have obvious FR problems you don't like, just use PEQ, and as little as possible. Or correct via the crossover if possible/you can do it. That's my conclusion.

Others just do what you like. No rules here. If you like RC, use it!

 

[morricab]

Literally as I said. It was a free PC demo version, and early one, s/w only. I used it to explore the effect of RC and FR corrections made by the software. I only played with the FR slopes a bit by dragging the plotted curves.

Here's what I feel about the room correction thing.

When I play my acoustic guitar, it sounds different in every room I play it. The interaction of the acoustic with the room is an entirely natural thing.

When studio music is recorded, a lot of care is used to eliminate room effects and many instruments etc are plugged directly into some studio hardware.

So doesn't it follow that I don't really need RC since room interaction is an entirely natural process for real instruments? I think so. And I hear so.

DIRAC is good at removing room artifacts but in the process strips the music of life and vitality. Such that it just sounds like it is coming from the speakers. It's clever, I don't deny it, but another issue is this. The maths processing used to do it is NOT transparent.

What I found when I was re-doing my Apogee crossover was this. I would use Peace/Equaliser APO to implement the notch I wanted to implement in the digital domain. I would model what I wanted by taking FR measurements and using XSim to work out what the circuit needed to be to effect the change. Those apps are just PC apps - Google them they are free and excellent.

Here's the crunch - when I prototyped the circuit in hardware using cheap components to start with, the hardware implementation sounded a lot better than the digitally applied PEQ.

The more PEQ you apply, the less transparent it becomes. That's just FR correction, though. Add RC processing in there too and that's a lot of mathematical mangling going on which has adverse effects on SQ. It's just the way it is with digital filters, I am afraid, though some implementations will be better than others.

This was all done through the PC, right? Not a hardware digital XO solution?

 

[Zero000]

This was all done through the PC, right? Not a hardware digital XO solution?

Yes. Just digital manipulation before hitting my DAC.

And why not if you only listen to digital sources? Masses of processing power on tap for free. Only the very best digital xovers could get close I suspect.

 

[morricab]

Yes. Just digital manipulation before hitting my DAC.

And why not if you only listen to digital sources? Masses of processing power on tap for free. Only the very best digital xovers could get close I suspect.

If the processor is dedicated it can be made to perform better than a general processor. Chips like the SHARC and other FPGAs have huge processing power for dedicated tasks. I am not sure it is so clear cut.

 

[morricab]

Yes. Just digital manipulation before hitting my DAC.

And why not if you only listen to digital sources? Masses of processing power on tap for free. Only the very best digital xovers could get close I suspect.

Also, if you didn’t control the jitter that could be horrendous from a PC. Finally, if your PC is using enough bits for the processing it shouldn’t have any audible degradation.

I also have the Behringer DEQ24/96 that when used as a digital equalizer digital in/digital out (sadly the DCX 2496 doesn’t have this) and jitter controlled was completely transparent. Despite its low price it has powerful processing.

 

[Zero000]

If the processor is dedicated it can be made to perform better than a general processor. Chips like the SHARC and other FPGAs have huge processing power for dedicated tasks. I am not sure it is so clear cut.

Well you can use DSP chips if you're feeling anal I guess.

I maintain DSP filters sound odd. Especially guitar effects ones trying to emulate valve amps. I know I have owned a couple. They have got better as the algorithms have gotten better but it is a tough problem to do well.

As room correction algorithms are.

 

[Zero000]

Also, if you didn’t control the jitter that could be horrendous from a PC. Finally, if your PC is using enough bits for the processing it shouldn’t have any audible degradation.

I also have the Behringer DEQ24/96 that when used as a digital equalizer digital in/digital out (sadly the DCX 2496 doesn’t have this) and jitter controlled was completely transparent. Despite its low price it has powerful processing.

Digital filters are not transparent no matter how many bits you use.

Jitter is a hardware issue and the are many solutions to reduce PC USB jitter.

 

[jasond]

A really great tool is https://www.focusfidelity.com/download consisting of the measurement tool Impala which is completely free and the Focus Fidelity Designer. It offers a much easier way than for example using REW and has other advantages too.
Its performance/result was great when creating convolution filters for ROON or HQ Player.