ASRC brickwall is also a possible cause, then. Would depend on the details of the post-ASRC 96 -> 48 and 48 -> 96 transitions in the chains you mention. This is OT to rePhase so perhaps we should continue discussion in PM or a different thread (pos, have you a preference?).
SigmaDSP parts share implementation with the AD1896 so the latter's datasheet offers substantial insight into behaviour of the ADAU1701 and ADAU1452 (among others).
SigmaDSP parts share implementation with the AD1896 so the latter's datasheet offers substantial insight into behaviour of the ADAU1701 and ADAU1452 (among others).
Phase better but sounds wierd
Im sure I screwed something up. Trying to correct my desktop listening position. Source is an Oppo 103D to a MiniDSP OpenDRC DA8 to a Yamaha power amp to wharfedale diamond 220s. Also have a sunfire mkIV sub, but that's off for these adjustments.
I corrected frequency response at the listening position and am happy with the results considering all the room issues (hard surfaces, sitting on a desk). Thought I would move onto to try phase correction. Here's what I did (mostly followed mwillems' tutorial: Guide to Speaker/Room Correction Using Freeware and JRiver):
Took measurements in HOLM of right and left channel at listening position (near field ~ 3 feet, laptop output HDMI to OPPO, UMK-1 mic). They were essentially equivalent so I loaded one into rephase. Choose a crossover linearization and port setting that got rid of wrapping. Used slide bars to get it pretty flat. Using 1650ish taps generated filters and uploaded them to the minidsp. Remeasurement shows phase no longer wraps in above used portion of frequency response.
But...
It sounds weird! I'm no golden ear so bear with me - soundstage is more open, more atmosphere, but everything sounds less natural. Voices too forward and less focused. I recalibrated without the slide bars - only trying to eliminate wrap, but didn't help.
Am I just not used to in-phase music
? What is the expected improvement?
Is it the single measurement at the listening position rather than close speaker measurements?
Thanks!
Im sure I screwed something up. Trying to correct my desktop listening position. Source is an Oppo 103D to a MiniDSP OpenDRC DA8 to a Yamaha power amp to wharfedale diamond 220s. Also have a sunfire mkIV sub, but that's off for these adjustments.
I corrected frequency response at the listening position and am happy with the results considering all the room issues (hard surfaces, sitting on a desk). Thought I would move onto to try phase correction. Here's what I did (mostly followed mwillems' tutorial: Guide to Speaker/Room Correction Using Freeware and JRiver):
Took measurements in HOLM of right and left channel at listening position (near field ~ 3 feet, laptop output HDMI to OPPO, UMK-1 mic). They were essentially equivalent so I loaded one into rephase. Choose a crossover linearization and port setting that got rid of wrapping. Used slide bars to get it pretty flat. Using 1650ish taps generated filters and uploaded them to the minidsp. Remeasurement shows phase no longer wraps in above used portion of frequency response.
But...
It sounds weird! I'm no golden ear so bear with me - soundstage is more open, more atmosphere, but everything sounds less natural. Voices too forward and less focused. I recalibrated without the slide bars - only trying to eliminate wrap, but didn't help.
Am I just not used to in-phase music
? What is the expected improvement?Is it the single measurement at the listening position rather than close speaker measurements?
Thanks!
Attachments
Last edited:
I did not see the cross post and started responding here: https://www.minidsp.com/forum/rephase/11626-phase-improved-but-sounds-weird#22804
So let's continue there if you don't mind
So let's continue there if you don't mind
I'll reply here. If you use some heavy smoothing before exporting from HOLM, you get a clearer phase curve. 1/2 octave works for me, sometimes even 1 octave. That lets you see phase tendencies more clearly and know what to correct of not correct. You don't want to correct every little kink, as POS pointed out on the other forum.
On the edge,
I've just tried rePhase under Windows 10.
Miraculously,it's running nicely.
(otherwise,stay under windows 7 as long as you can)
I've just tried rePhase under Windows 10.
Miraculously,it's running nicely.
(otherwise,stay under windows 7 as long as you can)
An externally hosted image should be here but it was not working when we last tested it.
Hello Thomas,
I was thinking to a minor improvement,
The possibility to export impulse as Frequency.txt ( Freq,amplitude,phase columns).
As you are generating impulse by iteration time/frequ. domain,maybe it's "easy" to add.
It's doable easyly with third-party software,(import impulse--->export as frequency.txt).
Like Holm,REW...
In fact,"just" an FFT of the impulse.
That allows to import directly a target response from measurement software.(with insane slope and phase response...)
I was thinking to a minor improvement,
The possibility to export impulse as Frequency.txt ( Freq,amplitude,phase columns).
As you are generating impulse by iteration time/frequ. domain,maybe it's "easy" to add.
It's doable easyly with third-party software,(import impulse--->export as frequency.txt).
Like Holm,REW...
In fact,"just" an FFT of the impulse.
That allows to import directly a target response from measurement software.(with insane slope and phase response...
)
Last edited:
folks,
I am following the step by step instruction from the miniDSP tutorial, i assume the procedure is the same even if JRIVER is used as the convolution engine.
Tweeter and Mid filters are easy. But the woofer measurements are near field, any advice on how to use that correctly?
any pointers in appreciated
I am following the step by step instruction from the miniDSP tutorial, i assume the procedure is the same even if JRIVER is used as the convolution engine.
Tweeter and Mid filters are easy. But the woofer measurements are near field, any advice on how to use that correctly?
any pointers in appreciated
Depending on the upper frequency limit and woofer diameter you might get away with only a close mic measurement for the woofer. Then you might also have to deal with baffle step...
But wait, what are you aiming at here? Are you correcting an existing speaker, or designing a full active crossover in FIR?
But wait, what are you aiming at here? Are you correcting an existing speaker, or designing a full active crossover in FIR?
The parallel W22s cover from 300Hz down to 60Hz.
XXLS from 20Hz to 60Hz
Will also use shelving to overlap the W22s and XXLS between 60Hz and 100Hz
Should i use the NF curves for them? Since they are in parallel, i will also need to do level match against the fairly low sensitivity W15 covering 300Hz to 2KHz
XXLS from 20Hz to 60Hz
Will also use shelving to overlap the W22s and XXLS between 60Hz and 100Hz
Should i use the NF curves for them? Since they are in parallel, i will also need to do level match against the fairly low sensitivity W15 covering 300Hz to 2KHz
Do the speakers already exist?
The baffle should preferably large enough to push baffle step problems where it does not matter as much, and preferably outside of the range where you will want to crossover your drivers to simplify measurement and consistency.
Have you looked at the LS1 design from Grimm Audio? It uses the same woofer and a sub. The tweeter they are using allows for a lower crossover frequency with a very well controlled directivity, which would remove the need for the 15cm driver.
This is a very sound design and probably not too hard to replicate.
The baffle should preferably large enough to push baffle step problems where it does not matter as much, and preferably outside of the range where you will want to crossover your drivers to simplify measurement and consistency.
Have you looked at the LS1 design from Grimm Audio? It uses the same woofer and a sub. The tweeter they are using allows for a lower crossover frequency with a very well controlled directivity, which would remove the need for the 15cm driver.
This is a very sound design and probably not too hard to replicate.
pos,
this is the speaker
http://www.diyaudio.com/forums/multi-way/272781-advice-needed-4-way-loudspeaker-31.html#post4554641
this is the speaker
http://www.diyaudio.com/forums/multi-way/272781-advice-needed-4-way-loudspeaker-31.html#post4554641
On LS1 website LS1 | GrimmAudio the third point in download section offer a design concept whitepaper, especially section 2.2 and 2.3 have some good info about what to correct and not to correct with DSP steering.
Yes, that white paper is pretty enlightening, and this is partly what go me going writing this software.
That said, FIR is not limited to brutal impulse response inversion as shown in section 2.2, and the equivalent of the phase linearized IIR+FIR filter of the LS1 can also be done entirely in FIR, using the exact same techniques (manual corrections, etc.).
That said, FIR is not limited to brutal impulse response inversion as shown in section 2.2, and the equivalent of the phase linearized IIR+FIR filter of the LS1 can also be done entirely in FIR, using the exact same techniques (manual corrections, etc.).
yes
sounds like a fun project for some very nice speakers!
jojip, maybe this could be of some help for you, as you setup is similar to Giant's one:
http://www.diyaudio.com/forums/mult...ion-eq-fir-filtering-tool-68.html#post4322701
http://www.diyaudio.com/forums/mult...ion-eq-fir-filtering-tool-68.html#post4322701
Do any of you have the same problem that I do? Basically your living room turns into a spaghetti mess of cables and components when you go the DSP route?
Here's an interesting solution which I'd never considered until yesterday. Basically you buy ONE computer, and then you install all of the components into a series of virtual machines.
For instance, you could have a setup like this:
virtual machine #1 : media server (this VM would be connected to a monitor, and it would provide the interface for your music, movies, etc)
virtual machine #2 : DSP (this VM would be responsible for crossover filters, delay, etc)
virtual machine #3 : your desktop computer (this VM could be for everything that you want to keep secure. For instance, I don't want to give my teenage kids access to my banking accounts. A dedicated VM for my personal desktop walls them off from my desktop computer.)
virtual machine #4 : a Windows XP VM for that old software that doesn't run on Windows 7-10 (Akabak, I'm looking at you...)
It's clever I think; at the moment I have a desktop computer in the garage for measuring speakers, I have a PS3 that I use as a media server, I have a NAS, three iPads and six laptop computers. By going the VM route, you can consolidate most of that down to one machine. I'm already in the habit of controlling my desktop from my laptop; I use VNC to connect to my desktop computer in the garage, so the VM solution is a logical choice.
My "inspiration" came from some dude on Youtube who was doing this for LAN parties, but you could do it for Rephase too:
https://www.youtube.com/watch?v=LuJYMCbIbPk
The big breakthrough here is something called "passthru." Basically you can "pass through" components to the VMs now. For instance, if you purchased a $200 sound card for Rephase, you can "pass through" from the host OS (linux) to the guest OS (Windows.) This is especially handy if you have a pile of hardware installed; for instance the dude on Youtube managed to shoehorn seven video cards into a single system. You can pass those through too; for instance you could have one VM for your media server that's attached to one monitor, and another VM for your DSP that's attached to another monitor. These are completely independent systems, so you can happily fiddle with DSP settings while someone else is perusing the media server. They could even have independent sets of mice and keyboards.
Here's an interesting solution which I'd never considered until yesterday. Basically you buy ONE computer, and then you install all of the components into a series of virtual machines.
For instance, you could have a setup like this:
virtual machine #1 : media server (this VM would be connected to a monitor, and it would provide the interface for your music, movies, etc)
virtual machine #2 : DSP (this VM would be responsible for crossover filters, delay, etc)
virtual machine #3 : your desktop computer (this VM could be for everything that you want to keep secure. For instance, I don't want to give my teenage kids access to my banking accounts. A dedicated VM for my personal desktop walls them off from my desktop computer.)
virtual machine #4 : a Windows XP VM for that old software that doesn't run on Windows 7-10 (Akabak, I'm looking at you...)
It's clever I think; at the moment I have a desktop computer in the garage for measuring speakers, I have a PS3 that I use as a media server, I have a NAS, three iPads and six laptop computers. By going the VM route, you can consolidate most of that down to one machine. I'm already in the habit of controlling my desktop from my laptop; I use VNC to connect to my desktop computer in the garage, so the VM solution is a logical choice.
My "inspiration" came from some dude on Youtube who was doing this for LAN parties, but you could do it for Rephase too:
https://www.youtube.com/watch?v=LuJYMCbIbPk
The big breakthrough here is something called "passthru." Basically you can "pass through" components to the VMs now. For instance, if you purchased a $200 sound card for Rephase, you can "pass through" from the host OS (linux) to the guest OS (Windows.) This is especially handy if you have a pile of hardware installed; for instance the dude on Youtube managed to shoehorn seven video cards into a single system. You can pass those through too; for instance you could have one VM for your media server that's attached to one monitor, and another VM for your DSP that's attached to another monitor. These are completely independent systems, so you can happily fiddle with DSP settings while someone else is perusing the media server. They could even have independent sets of mice and keyboards.
Last edited: