DiyMobileAudio.com Car Stereo Forum banner

81 - 95 of 95 Posts

·
Registered
Joined
·
731 Posts
I think it may but I would still need two ports to do a loopback. MacBooks only have one port.


Sent from my iPhone using Tapatalk
 

·
Registered
Joined
·
4,185 Posts
Discussion Starter #82
So Tried a new one with the new old re calibrated sound card

The result is not worse. The center is super defined...The center steals the show by far, vocal is super real and placed amazingly good now. The stage depth is good....a little different than dirac... still need to tweak that a little, maybe a few more tries. stage placement is good. Midbass response is better but different....not sure yet what..... Its good....
I need to listen for a few days and pick it apart


overall. Its good. I'm happy. It works!

Heres a pic of the left ch in rew and rephrase and the right ch in rephrase ( didn't take screenshots of all the tabs... you should get the jist of it




 

·
Registered
Joined
·
4,185 Posts
Discussion Starter #83
So changed a little bit

This time I moved the excess phase the same amount on both sides equally under 2.6ms long (the width of car) which is the narrowest dimension of the car

What I didn’t like is now better and it sounds excellent.

I also added eq work to the right to better match the left and simply ignore the dips
The left side has a dip at 1k and the right side has dips at 150,630,1k,.....

It’s very close to a Dirac tune now.....
I love the freedom to do it how I choose

I can’t get over how awesome the center is.
Vocal sounds astonishing real. The far rear stage on left is still a tiny bit wonky. After driving to work today I think I need to lessen the minimum phase correction on the left or lower the Q on one of the bands.

I should post up the spectral responce , very nice very even

I’ll do that in a few after work
 

·
Registered
Joined
·
4,185 Posts
Discussion Starter #84
I got it working excellent now

I tried every combination of phase eq that I could possibly do that makes the excess phase as flat as possible while keeping the same settings on left and right channels

What was left was a little at 300-630-5000

It’s badass now. Stage is right, sounds great, I ignored the big dips and the dip a 1k on the left.

The music has more snap and the sharp transients rip oh so nice now.

I am working on screenshots now
 

·
Registered
Joined
·
4,185 Posts
Discussion Starter #87
I would say that's a pretty close match....They look very alike. Its sounding excellent!..

As good as dirac now........I need to listen for a few days to really be sure and switch back to my Dirac Tune and examine. Super stoked I'm finally getting it
 

·
Registered
Joined
·
4,185 Posts
Discussion Starter #88
the right spectrogram shows a little more meat in the 400-1k range than the left..That was a little more energy as its about 1db louder in that range to get the center pushed center instead of in front of me.....
It shows I have it "balanced right" about 1 click.... Crazy how much that graph reveals about the system and its linearity
 

·
Registered
Joined
·
4,185 Posts
Discussion Starter #89
It’s very good.....but I can’t beat Dirac....

After two days of listening to this attempt. (Number 10) maybe 11 will be the magic one for me ....we’ll see

As of now it’s not better than Dirac....it takes a long time to get it right but is very fun to try to beat Dirac. The sub bass I like better on my tune. But that’s it , the mids and highs just aren’t better yet.
 

·
Registered
Joined
·
4,185 Posts
Discussion Starter #90 (Edited)
Got a. Excellent sounding tune (15th iteration)

So for a car I’m thinking this works good so far , left and right transfer functions match very nicely now and the sound is very similar to Dirac ....except I was able to set the magnitude more to my taste and use live RTA (which in my opinion is the best sounding in a car)

So I used REW and got all speakers time aligned and ignored the fact that the imaging was off and waited.

Than I used live RTA pink PN with 8k bins
And set the amplitude from 20k to 1.3khz on the left channel to flat as possible with only the left channel playing. Than I applied the same settings to the right.

I basically did the left channel and worked that side only and applied everything to the right that I did on the left .

So working the left after using live RTA from 20k to 1.3k I than used sweep measurements. I basically copied the Dirac way of measurements in the box shape except I added 3 measurements at each point and each of the 3 measurements in the box were approximately 2” apart. So it was the normal 9 measurements with an additional 18 measurements so the box shape had 3 spots at each measurement point. That allowed me to get a better average at each spot, in case one of the spots had a funky spot.

That’s why I also used live RTA above 1.3k is because each measurement point would drastically change if I moved the mic even a tiny bit making the average non consistent.

Even though under 1.3k that phenomenon was not a problem I still added the 3 measurements at each point on the box just in case.

I carefully made the amplitude flat under 1.3k ignoring the dips in the average. (Btw rephase eq is badass it works so so good by far the smoothest responce I’ve ever hear in my car) I kept my Qs under 4. That was a big sound quality aspect most of all was keeping all my Qs low. They are mostly 1.7 to 2.3 and a coupe in the 3s and one is 4 using two 16band eq banks. The rephase eq can have up to 7 or 8, 16band banks of eq so the amplitude was made extremely flat.

It was more a process of using lots of wider Q eqs and wiggle the amplitude into being flat instead of trying to focus so much on making an exacting Minimil eq where is limited to let’s say 10bamds or less. Having all those eq banks really made it amazing.

I’ve never tuned like this but gosh dam it sounds so much better having a lot of eq bands all lower proportional Q or constant Q
I like proportional Q better I feel like the interpolation works better!

After that was done I took a new set of measurements this time only 3 at the center moving from back of my head to middle of my head to front of my head in a straight line
And applied a aggressive FDW (4cy) and cleaned up the excess phase on the left channel only. I used one one of the 3 and picked the one that had the flattest phase so I would do the least amount of phase EQ

Than I turned on the right channel with all the same exact eq as the left. Any peaks above 1.3k were brought down using real time RTA. Below 1.3 the only thing that was different was the modes (80hz was obviously louder than the left so was 150 and 300 and 630) I left those alone! I let them be louder. Those peaks need to be left.

I noticed leaving those louder not only keeps the center where it should be , but on the other side of car the left is doing the same thing. So the energy is there!

I took the 3 measurements on the right and applied the same FDW and cleaned up the excess phase.

Measured left and right and the IRs are very very close to the same. The two wraps are in the same spots and have the same general shape. The coherence is excellent. And the tonality is to die for.

No more poping sound from left midbass can turn it up nice n loud and the 80hz response sounds just fine. Not trying to adress the left door and just ignored it completely and worked around it works to me the best. I lost a bit of gain from working around that dip but the sum of left and right show a nice smooth response.

So when I measured left and right together the dips are still there but the right channel fills in a lot of those dips. Where the left is 10db deep the right is flat the sum is only a coupe dB of dip. Working then eq from the worst side (responce wise) made the easy side fill in and those room dominated frequencies you really can’t tell so much which side is playing and has the dip. The responce is there and that’s what matters.

You can’t change the room dimensions, so parts of that will be there no matter what. Working on the left side first knowing the right side didn’t have those issues again just made the center better. The added gain from those frequencies also made it to where I did not have to attenuate the left channel at all. Both left and right have the same output gain in dsp and left and right sound to me the same
 

·
Registered
Joined
·
1,516 Posts
That’s a lot to digest but I could see how that might be really nice. When I’m integrating subs into home theater, I also try to go with wider DSP changes to feather the audio instead of hitting it with narrow-Q hammers.

You said you applied an aggressor FDW (4cy). What do you mean there? What freq range is that?
 

·
Registered
Joined
·
4,185 Posts
Discussion Starter #92
Aggressive 4cy FDW

That means I’m windowing out almost everything, if you open rew and add a FDW 4cy window is eliminating almost all the reflections out of the responce. So the excess phase and minimum phase are very similar except a couple areas. That way your not over correction on phase eq which makes a whole bunch of new problems and the sq goes to crap real fast .

I’m only interested in areas that are not minimum phase and to make it minimum phase you can only move so much phase around. It’s easy to get lost I’ve found out the hard way.

I only want to see where the minimum phase and excess phase move apart from each other and I’ve only been moving no more than 15deg of phase at a time. And as little a 1 or 2 degrees of phase under 1khz with again wider Q phase changes as well. Smooth transitions seem to sound a lot better. And as long as the peaks are brought down
The small little ripples are just not worth chasing.....seems the more you try to make a measurement look good the worce and harsh your filters sound.

It’s so easy to get lost with rephase because you can work it to make your measurements look perfect, and than your like “hot dam this is going to be so spaceship “ than you load the filter and it sounds like complete ass.

So yeah I was at the point of frustration and I went back to basics and used all rules of tuning I learned and applied that. So I went with low Q filters because you can’t go wrong and lots of them and it finally is super smooth sounding.
 

·
Registered
Joined
·
4,185 Posts
Discussion Starter #94
So a few points I’ve sorta learned about all this that maybe is worth noting.

Phase is very position dependent.

It’s easy to get lost trying to chase down phase issues that simply are not there. Most of the time the measurements are not enough to do anything with as far as super fine tuning phase. So the thought of super fine tuning the phase is not a practical way to think and isn’t something that should be thought of as possible. The measurements simply are not good enough to try to get perfectly flat phase in a car.

When measurements are made , and time aligned and averaged you can get a good representation on the big phase issues and those are obvious and can be fixed easily and they sound great. Wraps that are in all the measurements are fixed easily with either a crossover linearization tool and some phase eq with the linearization tools.
Cabinet resonances can be fixed as well however the ringing from a cabinet or a dash pod or door pod can be fixed however the ringing artifact that causes coloration to the sound can not be fixed with dsp. That is something that a whole thread should be dedicated to. Cabinets that are of lower Q and have better damping obviously will color the sound the least and can be made to sound absolutely excellent with phase adjustments. High Q cabinets or pods can be tamed , however the dry, rough, and simply unpleasant ringing high Q pods and cabinets exhibit can simply be the death of any real sound quality and completely kill the balance between harmonics and the music will always have that dry harsh ringing. A true IB or low Q pod with thick walls and heavy damping is not a choice. It is the only way!

As far as excess phase, there’s a lot of excess phase. In a car so much early reflections makes up so much of what is heard it takes a lot of measurements to actually see what’s happening. Taking close measurements and working your way with measurements every 2” or so from the speakers to the seat will definitely help “see” what’s happening.

Using a FDW seems to me the only good way to get a good idea on what to correct.
A gate is not the ideal way to see the excess phase in a way that is meaningful. The excess phase will in most measurements be no where near the minimum phase. What I did was look at the minimum phase and look for areas that do not follow the magnitude, than look at the excess phase in that spot. The sound can’t break the laws of physics and can’t come out ahead of everything so it can only be lagging. So if a spot is ahead that really means everything else is lagging, which can be the case being that there is so much excess phase and a large portion of the sound energy has at least rubbed against something, maybe an underdash or a door or a window.

What I mean by rubbed against. The sound pressure may have had an increase due to 1/2,1/4,or 1/8th space loading. And the area where your head is the biggest open space in the car, so it goes from higher pressure to lower pressure. Down near your feet in the footwell and the bottom of the door there is the seat, a console, a firewall , a dash , and other obstacles for the air pressure molecules to find resistance. That resistance will speed things up. Imagine trying to fix phase issues of mids and highs that are mounted inside of a bass reflex port on a sub box.

This is where car audio ppl and the home audio ppl will simply have very different looking excess phase plots that don’t resemble anything like the minimum phase plot. Let’s think about what excess phase actually is.

You have measured phase, excess phase and minimum phase. All are a component of measured phase. The excess phase is only the difference between the minimum phase and the measured phase. There isn’t such thing as minimum or excess spl. How loud something was measured should therefore only be associated with the minimum phase.

This is why I think it’s very useful in a car to use the spectrograph and look at the energy content along with the waterfall and see how long something is hanging on. That’s another topic but still a very useful tool to help understand why and what when looking at minimum phase and excess phase plots.

Lastly, the home audio ppl like to say things like “don’t do this” or “doing this is only good for one listening position”. Be careful when understanding what is being talked about here and where we may only care about one seat and will only ever care about one seat and our rooms are not even possible to “walk around in”.

As far as one seat goes, I have (of course) drifted away from some of that only to try to get to the bottom of how to do this in a car and have had some good successes. I haven’t been able yet to fully say with great certainty because I’ve only been able to do this in one car. The one thing I’ve have good luck with in car deviating from the home audio norms is in regards to using phase eq to regain phase coherency after doing seperate left and right eq.

Basically what I did was work the left and right seperate as far as magnitude and phase wraps (crossover linearization) and the normal twisting things from a multi-way.
Than after left and right magnitude are flat use phase eq and make the measured phase match left and right. This has worked quite well actually and I know I’m experimenting into uncharted territory but I want to think that the measured phase plot at the center of my head (3 measurement averages) and use phase eq, iir all pass filters, fir all pass filters, and linearization tools and all the tools within rephase to make the measured phase plots match after doing left and right eq has worked good so far.

The only problem I have had so far was after making some phase adjustments some frequencies changed in amplitude and would require a few tries and some sober listening to see why the amplitude changes happened, some were a result of fighting crossovers as a result of the phase changes and some were results of different interactions between the room and the speakers. So it took a few tries to get it and it works pretty good.

I’m not of the opinion that it’s better than not doing seperate left and right eq , I have had much better coherence and a more accurate and proper sounding stage without the use of seperate left and right eq, however I’m starting to feel that above about 2k seperate left and right Eq with a post eq phase eq to make the left and right measured phase match is definitely better
 

·
Registered
Joined
·
4,185 Posts
Discussion Starter #95
Iteration 21 now......

And need to retract some of what I said in above post. Not a whole bunch , but had a discovery that was in front of me the whole time. So the GD and excess GD. That is all I was missing.

The excess GD really tells a lot so does GD.
I was blindly using rew averaging and impulse centering (which works good on a single measurement.....with HF in the measurement) duh. Silly me: I was letting rew center the impulse and the impulse looked centered on the IR window. So I was like , thinking that’s it right....

So the excess phase now is not that far off from minimum phase. I was needing to use excess GD and get the “all” flat areas that don’t have a bump or peak to be at 0ms line. It was a matter of using signal delay, and re shaping the crossover after to accommodate for that change so the crossovers sum correct again.

Than I was also thinking the peak is the beginning of the IR, forgetting about rise time! Oh yes rise time! I have heard ppl talk about it but never really cared about it I thought they were talking about the speakers transients that were factored into the design of the speaker. No no no no ...no I was wrong, rise time is how fast a speaker can start playing a given frequency. And midbass and subs obviously are before the peak of HF. I already knew that but never thought how I would have to consider that into any measurement. Yes. So the IR actually “starts” at the first sample any polarity movement is made in the IR. Not the peak.:....

So needless to say, I re did that entire thing. Luckily I saved the measurements before I changed anything in Rew so they were all Raw measurements.

Now I can repair the excess phase and get excellent sounding stage. My own rephase tune is finally as good as a Dirac tune. And I have a level of confidence now because I can see, and hear the results and it’s working.

Upon this realization I had tonight, I came across something strange. It appears to look like a LR4 is one cycle away on the LP side of a complimentary LR4 with the HP..... I am going to have to investigate this....I am probably wrong, but I triple checked everything and it come out the same every time. The wrapping from the Lr4 the way it wraps puts the HP to have exactly 1cy delay and the delay is made up from the “wraps of phase” caused by the filter. It seems to be and I’m just speculating at this point, but instead of using a fir filter to linearize the LR4 a 2nd order allpass IIR filter should be able to to remove that delay. So I’ll find out.
If my hunch is right, the IIr could be very powerful tool in this case
 
81 - 95 of 95 Posts
Top