[musicreo]

I just skipped through the video but I think all other aspects they are discussing are only important for real speakers. As we are transferring the equalisation to headphones we can even correct room modes and can make cheap speakers sound more like expensive speakers. However, we have all the uncertainties with the headphone equalisation.

 

[johnn29]

I've been doing lots of experimenting with virtual room correction and I like my results now combined with the experimental reverb management feature.

My real room has a ton of ringing in the low bass. I have rigid walls and floors, the benefit is crazy room gain (115db to 6hz on a small sealed sub -SB2000) but I get ringing.



RT60



See the above plot - that was only at 75db, the louder the worse the situation is. Bass response after real EQ is fairly flat, so multisub in the real room won't do anything to get rid of ringing from my understanding. See below (no smoothing):



All it can do is destructively interfere with the peak and kill it to flatten response. Still rings, but less peak. Besdies, that's another £~800ish, a ton of setup and pain etc.

To really get rid of it it'd take bass traps - which is a big project and will eat up room and I cannot be bothered.


Enter Impulcifer. I nailed a nice virtual room correction on top of what Audssey did up to 250hz. It killed some of the peaks I had in the midbass which you can hear on A vs B. But the ringing still remained. I get this weird sensation in my ears when I hear it too, not sure if it's some weird phasing thing but it definitely sounds jarring in my real room. A testament to the convolution method is I get a similar feeling with the headphones.

So I used the reverb management from the experiments folder to set everything to 300ms. No more ringing! I guess this is time domain management we can do in virtual rooms which are impossible in real world. What's interesting about the reverb control is when you take it to artificially low levels, like sub 200ms the bass perception is much lower. I guess the SPL is identical but tighter/dryer so it's perceived as less?

Jaakko - is it worth trying to roll this into the room targets for virtual room correction? i.e. an ideal decay time?

Thought it was worth sharing - an example of how you can improve on your real rooms with this stuff. Still blows my mind. And it's better than the Smyth - you cannot do any of this with the A16.

 

[jaakkopasanen]

Very interesting videos. I didn't know that the normal room measurement doesn't match what we hear. I believe this is 100% relevant for virtual room correction too.

 

[jaakkopasanen]

Jaakko - is it worth trying to roll this into the room targets for virtual room correction? i.e. an ideal decay time?

Roll reverb management into room targets. What does that mean? Adjusting the room target based on reverb time? Or having a decay time as parameter for Impulcifer?

 

[johnn29]

Only in the bass regions does what we measure on a simple steady state measure match our hearing. Through the transition frequency, which isn't a fixed point, it can help but after it - you are likely to make things either stay the same (at best) and at worse make things worse. I figured that since we place the mics exactly where our ear is it won't be a big issue for us though.

I figured the decay time should be a feature for Impulcifer proper (not experimental) and also a target decay time for virtual room correction. I've been watching a ton of videos here - it's between 250-350ms. We're not simply deadding the room - but it'll kill off any of the ringing. See here for more detail from Flyod Toole. You're killing ringing not reverb in the low frequency.

. Often reverberation times measured at low frequencies are the decays of a few under-damped room modes. This is not reverberation; this is ringing!

 

[jaakkopasanen]

I agree that reverb control needs to make its way from the experimental script to Impulcifer, and it will.

As I understood the problem with eqing stochastic range, havibd the mics in the exact same location doesn't fix everything because the measurements will be affected by reverb while our perception of frequency response is not. Please correct me if I am wrong.

 

[Joe Bloggs]

Often reverberation times measured at low frequencies are the decays of a few under-damped room modes. This is not reverberation; this is ringing!

You got me wondering when Floyd Toole joined head-fi...

 

[johnn29]

I agree that reverb control needs to make its way from the experimental script to Impulcifer, and it will.

As I understood the problem with eqing stochastic range, havibd the mics in the exact same location doesn't fix everything because the measurements will be affected by reverb while our perception of frequency response is not. Please correct me if I am wrong.

Actually that is true from the videos I watched. But when using Impulcifer to set a room correction limit I cannot even tell the difference between a 5k limit and full range. Like A vs Bing a track or movie there's no difference. Honestly I also struggle to notice the difference in a flattened frequency response, I just notice the tilt higher bass, lower treble. The only solution I can see is a limit on correction - unless you have something up your sleeve?

 

[jaakkopasanen]

Actually that is true from the videos I watched. But when using Impulcifer to set a room correction limit I cannot even tell the difference between a 5k limit and full range. Like A vs Bing a track or movie there's no difference. Honestly I also struggle to notice the difference in a flattened frequency response, I just notice the tilt higher bass, lower treble. The only solution I can see is a limit on correction - unless you have something up your sleeve?

I have some ideas but they might not work in practice. If I understood correctly that reverb doesn't affect the perceived FR, then doing measurements which exclude the reverb would be more accurate basis for room correction.

Convolving a very short exponential sine sweep with the impulse response, applying steep band pass tracking filter for this and the deconvolving it to a new anechoic impulse response might do the job. SNR is not an issue here because the convolved short sine sweep doesn't contain any noise or distortion because it's not an actual measurement so short sweep should be fine. Short sweep is important because that's way it's possible to have shorter octaves in the sweep (in time axis) and therefore a 100 dB/oct band pass filter cuts the reverb shorter.

Second thing I could do is to implement mixed phase filtering for room correction. This would mitigate some of the phase related problems in the stochastic range discussed in the video. Room measurement mic placement could help with this too.

Of course I have no idea if it's actually possible to pull these off...

 

[musicreo]

. Room measurement mic placement could help with this too.

If I understood correct the angle of the mic changes the recieved FR at higher frequencies when you measure in the near field. When you measure in the far field
the angle of the mic will change the recieved the FR only minimal as you can not distiguish anymore between reverb and direct sound.
From my amateur perspective I would say that all the measurements should be performed not too far away from the source.

 

[castleofargh]

From my amateur perspective I would say that all the measurements should be performed not too far away from the source.

Then you measure the source instead of the sound at the listener's position. We tend to care about the latter more.

 

[musicreo]

Then you measure the source instead of the sound at the listener's position. We tend to care about the latter more.

I don't want to say measure 50 cm in front of the speaker with your measure mic. But do your complete measurement as close as possible to the speakers. Don't do measuremnets with 3 m distance where you have more reverb when you can do useful measurements at 1.5m where you have more direct sound unless you have an optimised room where it does not matter if you're measuring at 1 m or 3 m.

 

[johnn29]

I tried that - an ultra near field measurement. Works great for desktop/laptop/phone use because it feels more natural. But for two channel the reflections actually help the sound for me. So I prefer listening to music with my 1.5-2m HRIR's. Some speakers also don't respond well to near field listening - they are expecting to have the room reflections fill holes in the frequency response.

There's a very good discussion on it here:

Also thanks to Impulcifer I can't now help hearing the bass ringing and SBIR interference in my real room. I've got no real choice where I place my front channels due to room layout. But at least I have a virtual fix and it doesn't matter so much for movies

 

[johnn29]

I've been playing with applying EQ in Peace with the HRIR in place to good effect. In my real system I found after playing with Denon/Marantz "Cinema EQ" feature I liked it a lot - it's a high treble roll off. It tames overly bright tracks and especially helps with movies for things like breaking glass etc.

I found a graph of the THX-Re-eq filter which is similar so copied that.

Think I've realized once you have the virtual room correction done, reduced bass ringing by clipping the reverb you can still play with regular EQ to improve to taste.

 

[johnn29]

Over on the Equalizer APO Discussion forums, Mutt, has created some great upmixing scripts based on Dolby Pro Logic in all its varients.

https://github.com/Dogway/emulation-random/tree/master/EqualizerAPO/Surround

So now we basically have a virtual AVR for Impulcifer - makes it really useful when watching 2 channel videos on YouTube and the like. Or improving imagining with a center speaker. I can confirm they work as supposed to with various Dolby Prologic test videos.