[yelnatsch517]

Has anyone tried using rePhase to build an active speaker? http://www.minidsp.com/applications/advanced-tools/rephase-fir-tool
I'm thinking something along the lines of JRiver + Acourate/Audiolense/Dirac + rePhase

I don't think I've ever seen posts on linear phase systems nor mentioned rePhase. Is it not important in regards to speaker design?
How does this effect group delay and phase delay in a system?

Just another question. How would someone get 21+ channels out of a computer and into a DAC?
3 way speakers x 7 speakers = 21 channels

 

[3ll3d00d]

The motu 24 can do 24 channels in one box - http://www.motu.com/products/avb/24ai-24ao

Otherwise you just need something that can sync multiple devices to the same clock, various pro audio device can do this.

I use Acourate in jriver btw.

 

[splotten]

Has anyone tried using rePhase to build an active speaker? http://www.minidsp.com/applications/advanced-tools/rephase-fir-tool
I'm thinking something along the lines of JRiver + Acourate/Audiolense/Dirac + rePhase

I don't think I've ever seen posts on linear phase systems nor mentioned rePhase. Is it not important in regards to speaker design?
How does this effect group delay and phase delay in a system?

Just another question. How would someone get 21+ channels out of a computer and into a DAC?
3 way speakers x 7 speakers = 21 channels

The developer of rephase has a thread over at diyaudio.com. Maybe they can help.

http://www.diyaudio.com/forums/mult...hase-linearization-eq-fir-filtering-tool.html

 

[Tip24/96]

Why would you need Acurate, or Audiolense if you are running JRiver with this rePhase system? Isn't Acourate and Audiolense just a method of software based DSP/EQ correction with FIR filters and such? What can Acourate and Audiolense do that can not be done with JRiver? I honestly don't know, that is why I am asking, plus I am truely curious.

 

[Tip24/96]

The rePhase tool from MiniDsp looks pretty cool, but, I am not sure how much phase, be it linear or not, matters in a crossover. Apparently you can hook one of MiniDsp's MiniSharks or OpenDRC units with rePhase and correct for any non linear phase issues in even passive crossovers.

How difficult would it be to design an active crossover with a MiniDsp OpenDRC using this rePhase software and one of their UMIK-1 mics? Would the delay in the initial impulse response mentioned be able to be corrected in the OpenDRC's programs?

 

[3ll3d00d]

Why would you need Acurate, or Audiolense if you are running JRiver with this rePhase system? Isn't Acourate and Audiolense just a method of software based DSP/EQ correction with FIR filters and such? What can Acourate and Audiolense do that can not be done with JRiver? I honestly don't know, that is why I am asking, plus I am truely curious.

acourate or audiolense calculate the filter(s) for you based on your measurements & various controls they let you tweak, with rephase you have to do it all yourself (and I don't believe it has the same features as acourate/audiolense either). RePhase is essentially a manual PEQ solution for phase, acourate/audiolense are semi automated speaker & room correction systems. The only thing they really have in common is that they generate quite long FIR filters.

 

[yelnatsch517]

The only thing they really have in common is that they generate quite long FIR filters.

Do you know how that would affect timing or latency issues during video playback?

 

[Nate Hansen]

Do you know how that would affect timing or latency issues during video playback?

That depends on the resolution of your filter. A filter that works well at sub-bass fequencies will be quite long and in my experience lip sync is an issue.

JRiver can be used to host the convolution files created by Accuorate/Audiolense/RePhase but I don't see any utility in using rePhase with one of the other packages. Active crossovers can be built in rephase and the channel routing needs to be done in a convolution configuration file, or the xo/eq can be done in JRiver's PEQ like I do and you can just use rePhase for phase correction.

The audibility of phase in regards to crossovers has been debated much, though most of the known speaker designers I'm aware of don't place much importance on linear phase crossovers. There are facets of speaker design that are much more important to get correct IMO. Lin phase is icing on the cake if audible at all. Many claim that it is audible though I haven't honestly looked into it that much.

 

[3ll3d00d]

How does this effect group delay and phase delay in a system?

it depends on what sort of correction you do but clearly if you build a filter that adjusts phase then it will naturally affect the phase response and hence group delay (given that group delay is a derivative of the phase response)

in case it's not clear from Nate's reply, jriver is the only video playback engine that is "dsp delay aware" AFAIK, i.e. it knows the delay incurred by your audio playback chain and accounts for this during video playback hence lipsync is not an issue at all. If you use an external device then you need to take care with the length of your filters otherwise you will have lipsync issues (as the ability to delay video is uncommon to say the least). This will lead you towards a minimum phase correction (traditional PEQ) at the low end reserving phase correction for the high(er) end. A more advanced/complex approach can be implemented by manually trimming/windowing the impulse of the filter so as to make the filter shorter but I'd reserve that for some point in the future.

Also as Nate says, audibility of phase is a debated subject but there is certainly a school of thought that says correcting the phase response of different speakers so as to target a consistent phase response across those speakers is a good thing (tm). At least one rationale being that hearing uses interaural phase differences for directional cues upto about 1500Hz hence a good phase correction should be able to produce a marked improvement in the quality of the soundstage.

This is certainly my experience anyway, there is a wonderful sense of space/cohesion to the soundstage (*insert cliches re walls disappearing here*).

I have also linearised a passive crossover but found it had v little, if any, impact. I have not, yet, tried a linear phase crossover on an active main speaker (though I do use a linear phase XO from mains to sub).

 

[desertdome]

Why would you need Acurate, or Audiolense if you are running JRiver with this rePhase system? Isn't Acourate and Audiolense just a method of software based DSP/EQ correction with FIR filters and such? What can Acourate and Audiolense do that can not be done with JRiver? I honestly don't know, that is why I am asking, plus I am truely curious.

I use Audiolense with JRiver. It can take measurements of all speakers sequentially in just a few minutes, it knows the exact delay required for each speaker/subwoofer down to the 100th of a millisecond, it allows one to set frequency dependent windowing so that you only correct the direct sound, it lets you design individual targets for each speaker/subwoofer, it uses proprietary crossover design that is unrelated to traditional topologies (Bessel, Butterworth, L-R), it can handle routing of 17 speakers and 8 subwoofers with almost unlimited channels to be used by those speakers and subwoofers, it allows for variations in bass routing and mixing that aren't easily done elsewhere, and it sets all speakers to the same volume level. It really is an amazing program.

The motu 24 can do 24 channels in one box - http://www.motu.com/products/avb/24ai-24ao

After trying a myriad of pro audio devices, the new MOTU devices are the best for home theater use. I put up a review of my MOTU 1248 in the Official Motu Pro Audio Interfaces thread.

 

[LTD02]

i almost can't believe this, but it appears that the pioneer mcacc "pro" has phase correction.

and from what i have read in the rest of their documentation, the guys at pioneer seem to have the right philosophy with respect to eq'ing room problems (fix the speaker/fix the bass/leave the rest alone/allow for gentle wideband eq by the user).

http://www.pioneerelectronics.com/ephox/StaticFiles/PUSA/Files/Home/2014_MCACC_Ref_Guide.pdf

mcacc pro is available in a $600 (list) receiver


http://www.pioneerelectronics.com/PUSA/Home/AV-Receivers/Pioneer+Receivers/VSX-1130-K


the only problem is that receiver doesn't have pre-outs.


but that can be solved for $20 per channel with this and most folks probably don't need outboard amps for surround channels once the mains have been taken out of the power loop:


http://www.mcmelectronics.com/product/555-15370


-20db would seem about perfect for converting speaker to line. frequency response looks great. distortion looks fine.

 

[Nate Hansen]

3ll3d00d nailed it sorry for not being clear. My personal use is pretty much exclusively streaming content which makes lip sync an issue.

 

[LTD02]

that other thread gave me a headache but it did get me thinking that while linear phase itself may not be of paramount importance, it does allow for running very steep/brick wall crossover filters and that might actually be where the benefit is.

 

[LTD02]

...a good phase correction should be able to produce a marked improvement in the quality of the soundstage.

This is certainly my experience anyway, there is a wonderful sense of space/cohesion to the soundstage (*insert cliches re walls disappearing here*).

I have also linearised a passive crossover but found it had v little, if any, impact. I have not, yet, tried a linear phase crossover on an active main speaker (though I do use a linear phase XO from mains to sub).

so you had a good result with linear phase in an active system, but not so much in linearizing a passive system?


if that is right, what slopes were you running on the active system and any other thoughts on why the delta?

 

[LTD02]

was curious about the delay added by this kind of correction.

would it be correct to assume that if the top/mid is crossed over 4th order and the mid/bass is crossed over 4th order, we have 720 degrees (roughly) of phase 'roll' (2 complete rotations), so to put sufficient delay on the tops in order to allow the bass to 'catch up', the tops would have to be delayed by twice the period of the bass, which for a 50hz tone (20ms per cycle) would be 40ms?
is there a rule of thumb for where this becomes too much?
i see on the Yamaha pro powered p.a. speakers, they are only doing linear phase on the mid/tops. i take it this is because time aligning the bass would create too much delay for live performances (which of course have no video delay component)?


just trying to get my head around some of the fundamentals.

 

[3ll3d00d]

so you had a good result with linear phase in an active system, but not so much in linearizing a passive system?

not quite. What I meant was that I could detect no difference between a typical acourate room correction (which does a variety of things including excess phase correction as measured at the listening position) vs that same room correction approach applied to a passive loudspeaker which has been measured quasi anechoically & an excess phase correction applied such that the measurement sweep contains that phase correction. The most obvious artefact corrected was a phase wrap around the passive XO frequency (1500Hz), this graph shows the corrected and uncorrected phase response. Brown is the native phase response, green is the corrected one. Measurement taken at ~1.2m with a 6 cycle frequency dependent window applied (I get 5-6ms of clean data in my lounge when I get the speaker/mic in the right place so a 6 cycle FDW is about right).

In order to determine whether the passive crossover excess phase correction in itself was useful, I'd probably have to just compare them without any correction vs with just the excess phase correction of the crossover. That is a bit of academic exercise really so I haven't done that (as this isn't a job!)

if that is right, what slopes were you running on the active system

the active part of the system is just my sub really, I use a 2nd order Neville Thiele crossover. Here's a pic to compare an NT2 vs an LR4; red/green is the NT, blue/brown is the LR.

 

[3ll3d00d]

is there a rule of thumb for where this becomes too much?

too much for what? lipsync with AV? if so something like 40ms is the outer limit I think, preferably more like 25ms at most though

you can play with correction in rephase to see what sort of impulses result to see how long the resulting filters are

 

[LTD02]

thank you. i just read bruggermann's white paper 'thoughts about crossovers'.