DiyMobileAudio.com Car Stereo Forum banner

161 - 180 of 928 Posts

·
Registered
Joined
·
732 Posts
Also interested to know if they use FIR filters. I know this DSP is capable of it.


Sent from my iPhone using Tapatalk
 

·
Premium Member
Joined
·
2,418 Posts
Also interested to know if they use FIR filters. I know this DSP is capable of it.


Sent from my iPhone using Tapatalk


Yes it uses some FIR and some IIR filters.


Sent from my iPhone using Tapatalk
 

·
Registered
Joined
·
4,304 Posts
Discussion Starter #163 (Edited)
I really haven't done much research on DIRAC and do not own a MiniDSP or any other piece of equipment that incorporates the technology. So, please help me to understand.

I thought the DIRAC measurements and algorithm process generated a combination of multiple IIR and FIR filters (aka mixed-phase) to correct magnitude and impulse response into a single "filter" that the signal is passed through. While it might be a single "filter" by name, isn't this really just a package or delivery method for an unknown combination of multiple IIR and FIR filters? If so, then I doubt you'll have much luck getting MECA R&E to approve the DSP for Stock Class.


It was half hearted. Yeah I don’t compete and don’t care to.
But I’m certain it wouldn’t qualify as a single filter unless there was the inevitable proof.

Yeah, if anyone has ever worked with minidsp , than you know you can make an advanced filter set that is basically just a string of biquads that if long enough would take up many “peq” banks or the like. With biquads you could write some very unique and uncommon filters to achieve different things that aren’t exactly peq. Like a inverse magnitude filter is what I am leaning towards with Dirac, but only after making it invertible using an fir, while ignoring parts that it sees shouldn’t be messsd with (high Q dips , etc) which if you made a inverse mag filter and included high q dips it could severely damage the speakers. So they did it right no doubt. A lot of thought definitely went into this.

It would be worth a look for the board to define this topology in the arena.
 

·
Registered
Joined
·
4,304 Posts
Discussion Starter #164 (Edited)
Do you think that possibly all this work you're doing for us about finding good mic positions may only be useful to those with horns because of the massively different dispersion patterns they have vs cones?

Also, You said something in the video about making sure your levels were close to each other in REW and not to use the levels in DLC to make everything close, otherwise it'll sound weird. What do you mean by that? Do the levels not stay where you put them, because i have to adjust them individually to get them to play within the green box.

Also, would you consider making a video that specifically shows where you measure. You were talking about using sofa mode and having the passenger side be incorporated..which is fine but there is 4 measurements on the left side that are hard to kinda contemplate of where to put them. I think sharing your exact mic positions would be super helpful :)

Sent from my XT1710-02 using Tapatalk
I would love to make another video.

But let’s be clear gang, my measurement points are merely what works in my car, I know car acoustics enough to know the freakness of car acoustics to say that everyone should find there own “pattern” of mic positions. Ya know the chair and sofa are the same algo , it just recommendations on where to put the mic for different types of seating. You can change it to what you want, just keep the right and left integrity between measurement points. And some common sense , try to be somewhat symmetrical, but definitely deviate a little.

Keep in mind that with 2.0 that is very different and the measurement points need to be like picture with 1.7 (what we use). You can and should put them anywhere. The reason is in 2.0 it corrects for comb filters between stereo pairs so it would have to have certain parameters to rephase the drivers for better stereophony. In our version it makes the transfer functions match. (Room correction and eq


I was talking about let’s say your tweets were 10db louder than mids or subs 20db louder than mids , use an rta and set gains so the acoustical shape of magnitude is in line with how you want to draw your target, dosent have to be perfect but the tighter it is the more gain you’ll get so it dosent have to cut something down to another pairs level.
 

·
Registered
Joined
·
4,304 Posts
Discussion Starter #165 (Edited)
Also interested to know if they use FIR filters. I know this DSP is capable of it.


Sent from my iPhone using Tapatalk

It does in fact use fir

It Hs the analog devices ADSP21xxx chipset , it’s a sharc. That means fir capability. It’s been disabled till now with Dirac upgrades.

Lot has the ability of 8192 taps divided between 8ch at 48k sample speed.

That’s roughly 26ms time window for fir. So it can convolve the time domain up to 26ms per stereo output. (4 left channels and 4 right channels ) or however you run your mixer. , you could if wanted to have a 52ms time window on a single channel, if you only wanted to Dirac the sub. Don’t know how it would handle it. But you could if wanted to

A tweeter needs about 50taps at 48k
A mid needs about 200 taps at 48k
A midbass needs about 1500-2000 taps at 48k
A sub could use a lot more? But with access to time delay per channel and polarity switch it would not need much more than 1500-2000 as well with a crossover about 80hz or higher.

Even with a 60hz crossover and access o polarity and delay 2000 taps would suffice.


What I mean by that is if the sub and midbass are 140 degrees off from each other a polarity flip would make them only 30 degrees, than the convolution, which may put everything at -360deg. Same thing for delay ,
If the sub has 10ms of group delay over half the in band , convolve the higher half (less taps) and add 10ms delay to the high frequency channels and all back to normal with much much much less fir.


So that one sharc on board is plenty to get the job done
If anyone wants to post there after IR screenshots you’ll see the delay on the IR peak and how far it’s delayed from the before IR peak, that delay shows how much fir is being used. (My Dirac’s always use all available fir) but you could see for yourself.
 

·
Registered
Joined
·
3,278 Posts
So i got a small tune going. Just to see how this thing works.

Let me say this. Do. Not. Buy one of these if you don't have some heavy dsp experience already. Running between the minidsp software(which is a lot more complicated than the standard c-dsp) and the dirac software and getting curves from Jazzi companion. Its A Lot to deal with.


I definitely see what you mean with the high frequencies and measurement points. This is a going to be a process to learn.

Im curious on how high i should be doing the box above my head. Im gonna try standard chair and standard sofa. See how they sound and explore from there.

The center is close. Not quite, but close enough. Not sure how to try and change that with measurement points??

I am getting a Very annoying issues that i had with any version of dirac I've used and thats the windowing being absolutely trash in letting you control it. It shouldn't just stop at any orange points you set up. It should let you push past anything you eant. Also at the very edges there are these large spikes for no reason that i cant get rid of. Ill post a pic of it.

The stage feels pushed onto the hood. Thats good. The instruments are super clear. That's very good. Im not gonna say it beats my tune because at this point..i don't really like how complicated it feels but ill report later once i get a feel and try some different curves and get the HF toned down.


It has Great potential. Just a big learning curve.





Sent from my XT1710-02 using Tapatalk
 

·
Registered
Joined
·
4,304 Posts
Discussion Starter #167
So i got a small tune going. Just to see how this thing works.

Let me say this. Do. Not. Buy one of these if you don't have some heavy dsp experience already. Running between the minidsp software(which is a lot more complicated than the standard c-dsp) and the dirac software and getting curves from Jazzi companion. Its A Lot to deal with.


I definitely see what you mean with the high frequencies and measurement points. This is a going to be a process to learn.

Im curious on how high i should be doing the box above my head. Im gonna try standard chair and standard sofa. See how they sound and explore from there.

The center is close. Not quite, but close enough. Not sure how to try and change that with measurement points??

I am getting a Very annoying issues that i had with any version of dirac I've used and thats the windowing being absolutely trash in letting you control it. It shouldn't just stop at any orange points you set up. It should let you push past anything you eant. Also at the very edges there are these large spikes for no reason that i cant get rid of. Ill post a pic of it.

The stage feels pushed onto the hood. Thats good. The instruments are super clear. That's very good. Im not gonna say it beats my tune because at this point..i don't really like how complicated it feels but ill report later once i get a feel and try some different curves and get the HF toned down.


It has Great potential. Just a big learning curve.





Sent from my XT1710-02 using Tapatalk

Please don’t say things like “don’t buy one if”
It took me 1hr to get the hang of it. Ppl are smart , look if anyone can buy a new phone and get your email and browse the web than you have half a brain and can get this fairly easily.

It’s WAY more simple to use than a macaroni or a helix or anything else, it’s full color display and the best support pages and live help. And it walk you through the tuning process.


It’s not the dsp that ppl have a hard time with, or setting a crossover, it’s the measurements and getting good results. You do not need any tuning experience to work this thing. If you can draw a line your good.

I mean, ppl figure things out. If your in the market for a dsp , I promise this is the easiest to work out of any of them., even the auto target sounds good , or better than what most ppl do on there own.


Yeah there’s a lot of controls. But it’s easy. Configuration of my email servers is much harder and more technical, and most everyone can do that.

I know for a fact video editing is harder than this , managing multiple video feeds and an audio track and all the effects...uugh that’s hard. This is easy.
 

·
Registered
Joined
·
3,278 Posts
Please don’t say things like “don’t buy one if”

It took me 1hr to get the hang of it. Ppl are smart , look if anyone can buy a new phone and get your email and browse the web than you have half a brain and can get this fairly easily.



It’s WAY more simple to use than a macaroni or a helix or anything else, it’s full color display and the best support pages and live help. And it walk you through the tuning process.





It’s not the dsp that ppl have a hard time with, or setting a crossover, it’s the measurements and getting good results. You do not need any tuning experience to work this thing. If you can draw a line your good.



I mean, ppl figure things out. If your in the market for a dsp , I promise this is the easiest to work out of any of them., even the auto target sounds good , or better than what most ppl do on there own.





Yeah there’s a lot of controls. But it’s easy. Configuration of my email servers is much harder and more technical, and most everyone can do that.



I know for a fact video editing is harder than this , managing multiple video feeds and an audio track and all the effects...uugh that’s hard. This is easy.
Yea but you're kinda a genius man. We're all just kinda trying to keep up with you these days.:p

Sent from my XT1710-02 using Tapatalk
 

·
Registered
Joined
·
4,304 Posts
Discussion Starter #169
When I had speakers up high , I had issues with polarity post Dirac,

Do not be aftraid to add a little delay to a speaker of hit the polarity switch on different drivers or try different crossovers after Dirac runs.

It can’t tell how high your sound is, sometimes a incorrect acoustic polarity makes the stage rise. And I am all about what I like not what is “correct”.

Even with some post Dirac adjustments , your still doing better than not having it at all.
A delayed flat response is still a flat response, it’s not uncommon to have to do a little of this or that to get it to image the way you want after it runs.

Eventually you’ll get a pre tune and a pattern that nails it. Just keep experimenting.

For me , I just know turning down the left horn -1.8db and flipping my doors out of phase before Dirac nets me a perfect stage. (But I know where the polarity is at crossover on all my speakers) it’s going to make linked channels match.


Also tray linking mid and tweet to a single target of your center isn’t right .
 

·
Registered
Joined
·
4,304 Posts
Discussion Starter #170
Yea but you're kinda a genius man. We're all just kinda trying to keep up with you these days.:p

Sent from my XT1710-02 using Tapatalk

Please.

I’m a fat old man with white hair and man boobs

It’s just when everyone else is playing games or going to movies I’m reading
The aes papers or Some technical journal
 

·
Registered
Joined
·
3,278 Posts
Is there any way /excel shortcut to bring an entire curve down by a set amount, instead of calculating -x amount every time?

Sent from my XT1710-02 using Tapatalk
 

·
Registered
Joined
·
4,304 Posts
Discussion Starter #172
Is there any way /excel shortcut to bring an entire curve down by a set amount, instead of calculating -x amount every time?

Sent from my XT1710-02 using Tapatalk

Don’t turn your stereo up as loud?

Not sure what you mean?


Are you editing or creating targets by editing the target file?
 

·
Registered
Joined
·
3,278 Posts
Don’t turn your stereo up as loud?



Not sure what you mean?
Like if i want to import a curve to dirac. If the curve is +9db above my biggest dip. I'd want to bring my entire curve down by 9 db. Is there an easy way to do that?

Sent from my XT1710-02 using Tapatalk
 

·
Registered
Joined
·
4,304 Posts
Discussion Starter #175
Like if i want to import a curve to dirac. If the curve is +9db above my biggest dip. I'd want to bring my entire curve down by 9 db. Is there an easy way to do that?

Sent from my XT1710-02 using Tapatalk
Oh I see.

I’m going to open a target in notepad and see how data is delimited,
Maybe try and find an offset and make a macro in excel and save as “extension”
And see if works
 

·
Registered
Joined
·
3,278 Posts
Oh I see.



I’m going to open a target in notepad and see how data is delimited,

Maybe try and find an offset and make a macro in excel and save as “extension”

And see if works
It should be something simple like a copy paste from Word to Excel and then some formula to say -10 from all in this row.

I just don't remember how to do that..its been a longgg time since H.S. Excel classes

Sent from my XT1710-02 using Tapatalk
 

·
Premium Member
Joined
·
1,168 Posts

·
Registered
Joined
·
3,278 Posts
Oh I see.



I’m going to open a target in notepad and see how data is delimited,

Maybe try and find an offset and make a macro in excel and save as “extension”

And see if works
This is the spikes im talking about at the edges of windowing


Sent from my XT1710-02 using Tapatalk
 

·
Registered
Joined
·
3,278 Posts
I am seeing some pretty big bugs or what i assume are bugs. A few complete crashes of the software. The DLCT is suddenly not working for me. Its playing its test noise through my left midbass no matter what output i choose but i can play music and everything is fine.

My left and right center are switched..and i have no idea how that happened considering when DLCT was working and i did a few tunes with it. All speakers were what they wear supposed to be.

Sent from my XT1710-02 using Tapatalk
 
161 - 180 of 928 Posts
Top