Compensating for mains and sub delay

[ernestcarl]

"wavelet is cool, but I find it rather confusing for transient analiysis. I think it's forte is showing reflections speratly, which is not the subject here."

Yeah, I don't see this being the case... From what I've gathered, various papers that do discuss transient response (time-frequency) analysis actually say to move away from Fourier only mode, for example: https://www.pearl-hifi.com/06_Lit_A...equency_Analysis_Methods_and_Applications.pdf

Even the REW manual says the following:

"For a time-frequency plot it would be more useful if the trade-off between time and frequency resolution varied with frequency, using a constant octave fraction for frequency resolution rather than a constant number of Hz and so giving higher time resolution at high frequencies and lower at low frequencies. A wavelet transform can achieve that ... "

But here's the thing, the wavelet transform spect does require you to change the scale parameters quite a bit depending on what you want to focus on e.g. frequency range of, say, HF vs LF.

You can also see pre-delay in Fourier mode, of course, but the ideal settings between the two displays would not necessarily be the same.

 

[OCA]

BUT my nasty 417Hz resonance is now pre-delaying AND ringing.

Wait for my next video (I was gonna make it today but I found every excuse not to all day - I loath editing videos!), I guess I've finally nailed proper speaker phase corrrection!

 

[ernestcarl]

I can even go further by merging diferent FDW windows at this second step. here is the result for 35 cucles for example (instead(!) of the 15 one)

I know it looks nicer, however, I think most people "in the know" seem to pretty much discourage long window corrections above the room's transition frequency. For that, you need at least (quasi)anechoic measurements.

 

[OCA]

parameters are also undeclared. I guess those things don't matter now.

Firstly, these allpass filters interestingly were not ringing despite magnitude and Q values way above what I had experienced all these years. My trusted step graphs totally failed to reflect the lack of echo! And then I was too excited to have finally found a sign for pre-echo in the REW CWT graphs you shared (and I mentioned that I owe it to your post!) and just took screenshots in a hurry rather than save REW graphs the proper way showing the settings, etc.

But to break the ice and to clarify any misunderstanding, I had used exactly the same settings you used in your earlier post. Reason being, spectrograms are not among my favorite graphs and seeing that there's a steep learning curve, I copied directly from you who I believe to be one of the few audiophiles out there who knows what he's talking about. And of course all settings were the same for both graphs although I agree that I should've stated that in my post.

But rest assured I'm not here to cheat or impress anyone. I only enjoy to learn and enjoy even more to share!

Anyways, a breakthrough phase correction method is coming soon Dirac will not like this.

 

[ernestcarl]

Firstly, these allpass filters interestingly were not ringing despite magnitude and Q values way above what I had experienced all these years. My trusted step graphs totally failed to reflect the lack of echo! And then I was too excited to have finally found a sign for pre-echo in the REW CWT graphs you shared (and I mentioned that I owe it to your post!) and just took screenshots in a hurry rather than save REW graphs the proper way showing the settings, etc.

But to break the ice and to clarify any misunderstanding, I had used exactly the same settings you used in your earlier post. Reason being, spectrograms are not among my favorite graphs and seeing that there's a steep learning curve, I copied directly from you who I believe to be one of the few audiophiles out there who knows what he's talking about. And of course all settings were the same for both graphs although I agree that I should've stated that in my post.

But rest assured I'm not here to cheat or impress anyone. I only enjoy to learn and enjoy even more to share!

Anyways, a breakthrough phase correction method is coming soon Dirac will not like this.

That's fine... it's just I get tired of having to guess what parameters are used if it's not clearly stated or included in the legend e.g. 1/3 vs 1/48 smoothing, FDW applied or not

 

[OCA]

FDW applied or not

I was a big supporter of FDW but lately I almost never use it. Firstly, the target curves are not designed with FDW in mind and secondly and more importantly, decreasing window size by frequency in a linear fashion lately feels to me a bit too optimistic to reflect the room. It's the fastest empirical way to at least sort of take care of the reflections but neither you nor me will be even remotely satisfied with its results.

Below transient, you've to target anechoic woofer response of your particular speaker (you have said the same thing above!) and remove the room from it for best results. I can't anechoic measure my speakers but luckily, magazines have done it for me so I immitate curves from magazine photos. And in that respect, even REW's custom target curve settings fail to satisfy. Dr. Toole's curve come closest to the bass roll-offs of most speakers but I had to create my own speaker curve as my target curve to achieve minimum phase at the LP.

Calling it a room curve is the most commonly agreed misconception IMO, it's always the speaker's curve you need to try to achieve. Speaker is minimum phase (or we want it to be or else we're all wasting our time) and room might agree to be min phase at just one point (LP) with the right filters which will remove it from the speaker.

What do you think?

 

[dualazmak]

At least in my case, even whatever the theoretical and experimental approaches could be discussed/applied in digital level and/or analog line-level and/or SP high-level "signals", the room air sound (L&R together) hearing (and measured) at my listening position is always the ultimate interest/target in my own room acoustic environment (not anechoic, not physically symmetrical, having several furnitures and audio rig). My final goal is to enjoy listening music, not hearing the "sound".

Only if you would be interested, please find here #774 on my project thread the latest system setup as of August 3, 2023; you can find there the best tuned Fq response curve of room air sound at my listening position.

 

[ernestcarl]

I was a big supporter of FDW but lately I almost never use it.

I don't necessarily find myself using this to completely fit all room measurements to a set FR target curve.

Yet, I do find FDW to be quite handy to improve visualization...

Desk rear surrounds

Sometimes I use it to reduce the number of manual/automatic PEQs -- for example, instead of five filters I can cut it down to three or fewer.


Couch BM Center channel




However, one has to be aware that the results are highly dependent on the reflection-decay characteristics of the room as well as the software windowing algorithm.

Full, Early & Late Frequency Response (Couch BM Front Left Channel)

*more or less the same produced in a Genelec GRADE report


Most of the time, when using FDW, I switch between 3-15 cycles to remove or reduce unwanted phase rotations and raggedness thereby making the phase graphs more "legible". Often, however, within my own dampened room the un-smoothed or normal smoothing options may be adequate enough.

As for the actual LF and HF slope... I use my ears. Occasionally I will get bored and boost the bass -- other times, I feel irritable and over-sensitive so I cut the HF with a simple shelving filter. But as long as it's flattish enough and not overly boomy in the bass, my ears pretty much can tolerate a wide variety of resulting speaker "room curves".

 

[OCA]

At least in my case, even whatever the theoretical and experimental approaches could be discussed/applied in digital level and/or analog line-level and/or SP high-level "signals", the room air sound (L&R together) hearing (and measured) at my listening position is always the ultimate interest/target in my own room acoustic environment (not anechoic, not physically symmetrical, having several furnitures and audio rig). My final goal is to enjoy listening music, not hearing the "sound".

Only if you would be interested, please find here #774 on my project thread the latest system setup as of August 3, 2023; you can find there the best tuned Fq response curve of room air sound at my listening position.

Dualazmak san,

Tbh (and I spent quite some time to understand it), your system is so specialized and out of ordinary that although you have a point, noone here will dare even testing it because it all looks very system specific no matter how wonderful your cross correlation aligning of drivers will be.

You are using $20K worth of amps as pre-amps for your bass-mid-twitter drivers and then you have horn tweeters running in parallel with different amplification. Altough, I love the enthusiasm and beauty of your system math (I've an aeronatuical engineering degree), as @ernestcarl explained to you earlier, these are not things anyone here have difficulty doing with REW as a first step to every calibration attempt. Enlighten us instead on the advantages of using such diversified amplification, differences between horn and neodymnium tweeters, etc. I am sure, everyone here is very willing to listen

 

[dingdong3]

As for the actual LF and HF slope... I use my ears. Occasionally I will get bored and boost the bass -- other times, I feel irritable and over-sensitive so I cut the HF with a simple shelving filter. But as long as it's flattish enough and not overly boomy in the bass, my ears pretty much can tolerate a wide variety of resulting speaker "room curves".

I agree to this.

Also, for what it's worth, I think the room curve is already set in your room, especially in the frequency range above about 500Hz, and you should acoustically treat it with basstraps or diffusers, along with the right (best) speakers and listening positioning.
Then you'll have a good room curve naturally, and you can adjust the bass and treble according to your own preferences. It's not a good idea to try to fix it by looking at the response recorded by the microphone and applying FDW or smoothing to the already created room response.
While the possibilities and effectiveness of FIR filters are great, taking physical acoustic steps to improve your sound should be your first priority....
And the bass region from about 200Hz down to 20Hz, when properly room acoustic+subaligned, has very little deviation from my listening position and elsewhere when I raise or lower the eq.
If it's fine in one listening position, but thumps in another, it means the bass energy hasn't really dissipated and it's a potentially unpleasant experience.
In a DBA or multisub experience, not in VBA, this energy is actually dissipated. It doesn't get slightly better, it loses the room's roominess.
To see this accurately, you need to look at the Spectrogram, Decay, and Clarity are actually not reliable enough to use as a guide in small rooms, especially if the room acoustics are poor.

 

[OCA]

Most of the time, when using FDW, I switch between 3-15 cycles to remove or reduce unwanted phase rotations and raggedness thereby making the phase graphs more "legible".

I was stuck in the exact same mistake for months. Forget about FDW. By the time you have EP soft enough not to ring, it's not even helping the real problem (and sometimes it does help just to keep you get stuck in a non-trivial solution longer!). You don't correct phase, you correct excess phase as all you are after is min phase at the LP. And yes, you can't move walls without major ringing effects, but you can ask the walls "not" to treat your left and right speakers so differently

 

[OCA]

you should acoustically treat it with basstraps or diffusers,

Wasn't my intention but when you run a DRC channel you get many (hundreds) measurements asking for a filter. I've managed to satisfy most subscribers with one major exception : when there's proper room treatment (as in traps good enough to swallow low bass but not necessarily treat both speakers the same way), DRC never worked!

 

[dingdong3]

Wasn't my intention but when you run a DRC channel you get many (hundreds) measurements asking for a filter. I've managed to satisfy most subscribers with one major exception : when there's proper room treatment (as in traps good enough to swallow low bass but not necessarily treat both speakers the same way), DRC never worked!

I never said you or your subscribers were unhappy, I think you misunderstand.
I never said DRC doesn't work.
It's just about prioritization. It is also true that DRC must be accompanied by proper room acoustics for it to work as intended...

At frequencies below 200Hz, there will inevitably be room modes depending on the desk, size of the room, height, etc. and it's certainly better to apply EQ than not, even if it means sacrificing bass.
Frequencies above 500hz are a different story, though, and a little shelf filtering to suit your ears (everyone's HRTF is different. Judging this from a micro-recorded response is misleading, especially in a reflective room that is not room-acoustic).
Probably because the frequency region above 500hz is not controlled for early reflections when considering FDW or smoothing.
If you control the early reflections within 10ms down to -20 to -25db, you won't have this problem. It will naturally create a room curve without any EQ.

It's a slightly different story, but I don't know why people put so much meaning into the target curve, unless it's something like the Dolby Atmos reference studio spec, but I don't know why the average user would apply the Harman in-room target as is.
It's not a bad choice if you approach it in the sense of compensating for loudness averaging, but I think there's a "preference" involved in whether or not that's the right target.
This is something that needs to be listened to by ear, as bass and treble are determined by "preference" and the correct "FLAT" response depends on the individual's "HRTF" as mentioned above.
Adjust the slope of a few decibels in a particular octave.
This is by no means a one-size-fits-all formula, nor should it be.

 

[dualazmak]

Enlighten us instead on the advantages of using such diversified amplification, differences between horn and neodymnium tweeters, etc. I am sure, everyone here is very willing to listen

Thank you for your kind reply.

First, you would Please find here (on the project thread) and here (remote independent thread post) the Hyperlink Index of my project thread and some of my related posts in remote threads.

Also you can find rather detailed background and preceding history towards my multichannel multi-driver multi-way multi-amplifier stereo project: #030, #128, #178 on the project thread, and #015(remote thread).

Let me very briefly summarize the essence of motivation and advantages in my DSP-based multichannel multi-amplifier project as follows:

1. Complete elimination of LCR-passive-network and SP attenuators (which wastes almost all of the signal energy into heat) from SP system for direct and dedicated driving by each of the "carefully selected" amplifiers. (for much better and efficient drive of SP driver, especially better transient behaviors; each SP driver only receives the specific Fq zone signal.)

2. "Right person in right place" amplifier selections for SP drivers; sub-woofers, woofers, midrange squawkers, tweeters, super-tweeters.

Spoiler

You would please be reminded that even Greg Timbers uses "reasonable and budget" small integrated amplifier Pioneer Elite A-20 (low THD, high S/N, RCA unbalanced input, 30W+30W [20Hz-20kHz, THD0.05%, 8 ohms]) for ultra-high Fq compression drivers (super tweeters) in his extraordinary expensive multichannel stereo system with JBL EVEREST DD67000 which he himself designed and developed: please refer to my post here #435. My approach and policy in multichannel audio project is essentially very similar to Greg Timbers's.
https://positive-feedback.com/interviews/greg-timbers-jbl/

3. Fully digital DSP EKIO processing enabling flexible (even on-the-fly) XO/EQ/Group-Delay_(0.1msec precision)/Relative-Gain tunings.

4. Further safe and flexible on-the-fly relative gain (tone) controls by each of the HiFi integrated amplifiers.

5. The very highly efficient HiFi metal horn super-tweeters (Fostex T925A) can well support/supplement Beryllium dome tweeter, and enables flexible compensation for age dependent hearing decline in high Fq sound (over ca. 7 kHz) depending on the audiences. (ref. here #643)

I know well, even within ASR Forum, several people build/built and enjoy very similar DSP-based multichannel multi-SP-driver multi-amplifier fully active audio system...