well it's pretty straightforward if you know linux, if you don't know linux then most of the guide would be learning linuxperhaps something like daphile might fit the bill? not sure if that accepts audio inputs or not though
Don't know linux at all
.... but been known to learn Looked at daphile...as you suspected, didn't see the ability to accepts inputs, but maybe it does and i just missed it
Anyway, how about AVLinux, or KXStudio ?
All i want out of the box is audio...and both of these seem to be especially geared for handling audio IO.
well it's pretty straightforward if you know linux, if you don't know linux then most of the guide would be learning linux
that right there is the GOSPEL TRUTH
If you are running a more substantial computer with plenty of processing power would there be problem running both an audio box as you all are talking about and still being able to have a working linux computer running Suse linux? Is the fact that you are doing convolution and such require the complete cpu's attention or is it more of noise or something else which would make you need a separate small dedicated linux box? Is it just to have a separate audio system standalone that is driving this way of doing things?
OK. I'll stop and wait for guide
here goes nothing
http://www.diyaudio.com/forums/pc-based/302900-brutefir-dsp-pc-step-step.html#post4967403
post there if you have any questions, or post any requests for more vids
the last video is the cool one.
here goes nothing
http://www.diyaudio.com/forums/pc-based/302900-brutefir-dsp-pc-step-step.html#post4967403
post there if you have any questions, or post any requests for more vids
the last video is the cool one.
Great thanks the guide over there it really makes me now try a non GUI Linux box, but why i post about it here is because the Rephase correction part for me looks wrong way to correct a minimum phase device, i could be wrong in what sense from the movie then please correct me and forget note below
.It looks like pass-band for center speaker is corrected with linear phase (FIR) amplitude correction and if that's the case final corrected amplitude response will be alright and flat corrected but phase will still be reflecting same un-flat phase as before the correction. A non flat amplitude minimum phase device corrected with minimum phase (IRR) EQ will correct phase flat when we correct amplitude and actual if we could IRR EQ correct a speaker in outdoor environment flat from DC component to 300kHz or more it would be mostly linear phase inside 20Hz-20kHz audio band, IRR EQ depends how trustful is measurement and how much room is in the measurement but that's often visible in REW EQ window where we can see waterfall or IR below and if example there is a 6dB peak and we turn that down -6dB and no improvement happens then its probably room or diffraction and needs other care or tricks.
Last edited:
here goes nothing
http://www.diyaudio.com/forums/pc-based/302900-brutefir-dsp-pc-step-step.html#post4967403
post there if you have any questions, or post any requests for more vids
the last video is the cool one.
Hey 1201, WOW...very nice...many thanks !!!!
If I can make it to the last video I'm home free, familiar territory there
Looks like a fair amount of work on first two videos , but certainly within doable.
I'll do the same as BYRTT, try and segregate comments to the appropriate forum.
I did notice the same thing he did regarding the linear-phase generation.
My take about that is we really want the eq split into both linear and minimum phase components...if we can discern which individual eq's go into which bin.
Eqs that are for correcting speaker response should go in the minimum phase bin, because those filters will address both magnitude and phase correctly. Eq's that are addressing room response should go in the linear-phase bin because room response isn't min phase. That's just my understanding, maybe folks will correct or amplify it. The REW section on min phase is pretty helful here IMO.
Again, thanks a bunch for your generousity !
Great thanks the guide over there it really makes me now try a non GUI Linux box, but why i post about it here is because the Rephase correction part for me looks wrong way to correct a minimum phase device, i could be wrong in what sense from the movie then please correct me and forget note below
It looks like pass-band for center speaker is corrected with linear phase (FIR) amplitude correction and if that's the case final corrected amplitude response will be alright and flat corrected but phase will still be reflecting same un-flat phase as before the correction. A non flat amplitude minimum phase device corrected with minimum phase (IRR) EQ will correct phase flat when we correct amplitude and actual if we could IRR EQ correct a speaker in outdoor environment flat from DC component to 300kHz or more it would be mostly linear phase inside 20Hz-20kHz audio band, IRR EQ depends how trustful is measurement and how much room is in the measurement but that's often visible in REW EQ window where we can see waterfall or IR below and if example there is a 6dB peak and we turn that down -6dB and no improvement happens then its probably room or diffraction and needs other care or tricks.
I believe( and I may be wrong) that if you take a flat phase response and perform any IRR eq, you will twist the resulting phase and it will no longer be non flat. at least that is what rephase shows when I import a measurment without any phase data. Any IIR modifications also twist the phase.
maybe POS can give some input
Hey 1201, WOW...very nice...many thanks !!!!
If I can make it to the last video I'm home free, familiar territory there
Looks like a fair amount of work on first two videos , but certainly within doable.
I'll do the same as BYRTT, try and segregate comments to the appropriate forum.
I did notice the same thing he did regarding the linear-phase generation.
My take about that is we really want the eq split into both linear and minimum phase components...if we can discern which individual eq's go into which bin.
Eqs that are for correcting speaker response should go in the minimum phase bin, because those filters will address both magnitude and phase correctly. Eq's that are addressing room response should go in the linear-phase bin because room response isn't min phase. That's just my understanding, maybe folks will correct or amplify it. The REW section on min phase is pretty helful here IMO.
Again, thanks a bunch for your generousity !
You are very welcome man. actually I normally eq the low end with minimum phase and the high end with linear phase. Most lately though Ive just been doing IIR eq with FIR crossover.
the good thing about brutefir is that later if I want to linearize the phase at the listening position I can cascade a separate filter on top of the EQ filter just for the phase data- thanks to Rephase for giving us the ability to do phase independently of magnitude.
There have been some posts either earlier here or elsewhere that show that if you have a substantial computer, it will remain well into the aerobic zone doing this kind of audio processing. The problem though is that if you use its headroom for something else it may get distracted and not perform the audio processing in a timely fashion.
I think what is driving this command line linux is cost - a bare bones approach to a dedicated processing engine for real time audio; dedication also helping to reduce processing distractions.
But please save me from command line interfaces. They are great so long as you know the command but what if not or if you mistype? For the last 20 years or so I've celebrated the fact that I no longer need to use them.
I think a Linux application that hides Linux from me in everyday use but allows me to dig in when I need to would be wonderful!
the problem i found with not using a dedicated setup is achieving lowest latency.
if you are not using it for video then you dont have to use partitions in brutefir and you can get by but when you are using many partitions for the absolute lowest latency possible, anytime you start another program brutefir will overrun and abort.
thats why I use a dedicated pc
I believe( and I may be wrong) that if you take a flat phase response and perform any IRR eq, you will twist the resulting phase and it will no longer be non flat. at least that is what rephase shows when I import a measurment without any phase data. Any IIR modifications also twist the phase.
maybe POS can give some input
My take about it is room is IRR correctable as long as first reflection arrives direct (remember wesayso said to get good correction results at listening position we should work on get a clean IR there in have later reflections down or lower than -20dB point). Also think about a tweeter that in datasheet is flat down to say 1kHz is a IRR device but if we measure it with out any baffle (direct reflection) it will fall off depending own diameters direct reflection in area say 2-3kHz. Below Jeff Bagby spreadsheet show baffle and room (boundary reinforcement/pressurization gain) dependencies and think if those first reflections arrive direct in time we can export frd file taken from red curve and sum it to raw speaker drivers performance and correct rest of red curves peaks and dips with IRR EQ. Problem is modes from spreadsheet vary with position therefor 2 inch round overs or more plus smooth baffle transitions pays back, the vary with position thing plus real world living room furniture is probably why room response is told as non minimum phase, also as example in second spreadsheet below where same speaker now has position with same distance to three boundary gives so deep a resonance its seems impossible to have the IRR dynamic range needed. So if room boundary reinforcement plus pressurization gain gets to listening position as direct reflection and is IRR EQ corrected to sum flat with woofers we should get natural good band pass relative flat phase down in low frq area, and use FIR to correct XO and other of systems excess phase.
Attachments
Last edited:
For me, it's needing a small form, headless box, that can be mounted in a traditional amplifier rack. I use my system both indoors as hi-fi, and outside for party / DJ type use.
The miniDSP rack pictured below works great. But it allows a max 6144 taps on the sub channel, which is ok, but I wish I had more to work with. The linux box would allow that. (or the JR ID if input could be made to flow through it.
Attachments
Mark, couldn't you just build the needed and optimised correction on a Windows machine with JRiver's WMD driver and copy that processing chain to the end solution (the JR ID). That would be a simple and quick solution if you ask me. Provided the JRiver on the ID has all of the processing DSP of the regular version.
Ronald, I don't think I'm following what you're saying...
By processing chain, do you mean further processing in addition to ID's convolution (which is working well) ?
I'm still not seeing any way to get external media to flow through the ID.
Picture someone at a party wanting me to plug in their phone and play something from it..
If you're showing me how to do that, sorry I'm just not getting it ...unless maybe you mean build a small form windows box to replace the ID? Or stream from a 2nd windows PC using WDM?
By processing chain, do you mean further processing in addition to ID's convolution (which is working well) ?
I'm still not seeing any way to get external media to flow through the ID.
Picture someone at a party wanting me to plug in their phone and play something from it..
If you're showing me how to do that, sorry I'm just not getting it ...unless maybe you mean build a small form windows box to replace the ID? Or stream from a 2nd windows PC using WDM?
So the input is the only thing holding you back. I'd figure any other Windows PC with JRiver on it should be able to route it's sound to the ID. At least, that's what I expect. And with that working you should be able to use that PC's input either trough the WMD driver or using the Open Live option.
At least I expect that to work, it worked when I used a separate DVD player (long ago) that had DLNA capability. I could run trough the DVD player to my TV with JRiver.
I dropped that route long ago but remember it working like that. I'd expect their own ID should be able to do something similar.
https://wiki.jriver.com/index.php/DLNA#Using_DLNA_--_Finding_DLNA_Renderers_and_Servers
At least I expect that to work, it worked when I used a separate DVD player (long ago) that had DLNA capability. I could run trough the DVD player to my TV with JRiver.
I dropped that route long ago but remember it working like that. I'd expect their own ID should be able to do something similar.
https://wiki.jriver.com/index.php/DLNA#Using_DLNA_--_Finding_DLNA_Renderers_and_Servers
Last edited:
Yes, input is the only issue... and I'm with you now. I figure what you suggest will probably work inside, and was on the path of trying that until my focus switched to a linux box.
But trying to use a win PC to flow input to the ID is still mostly for curiosity's sake...I just can't see it working outdoors.
But trying to use a win PC to flow input to the ID is still mostly for curiosity's sake...I just can't see it working outdoors.
you could easily build something in shallow 2U rack case (one random example https://www.xcase.co.uk/collections...m-atx-short-double-door-aluminium-front-panel) that will be more than capable of doing the job. Get a decent matx mobo that has all the connectivity you need onboard and has an m2 slot, add a CPU & an m2 drive & some fans and away you go.
Great reading. I'm 100% up to date I read this entire thread. Wow what a ride
So, I got quite a bit from it all. I have a sorta simple question for y'all ; so if a phase measurement goes down and is falling (like in REW) , is that because frequency increases per second?
I mean, if frequency is rising (the bottom bar on graph) and phase is time, (side bars) than as more and more cycles take place in every second the length of time gets smaller between cycles. And that's the scale phase is displayed on rew injust plain phase so that's why it shows it's falling? I'm trying to grasp the relationship between time and frequency on graphs. More than just reading the graphs but understanding what's happening to the loudspeaker and room rather than how "good" a graph may appear.
Any reading or links ?
Thanks
So, I got quite a bit from it all. I have a sorta simple question for y'all ; so if a phase measurement goes down and is falling (like in REW) , is that because frequency increases per second?
I mean, if frequency is rising (the bottom bar on graph) and phase is time, (side bars) than as more and more cycles take place in every second the length of time gets smaller between cycles. And that's the scale phase is displayed on rew injust plain phase so that's why it shows it's falling? I'm trying to grasp the relationship between time and frequency on graphs. More than just reading the graphs but understanding what's happening to the loudspeaker and room rather than how "good" a graph may appear.
Any reading or links ?
Thanks
Last edited: