[71 dB]

I'd say the main principle with crossfeed is to not pay attention to how much the sound chances, but does the sound become more natural and pleasant to listen to. Don't assume the original sound to be even near perfect for headphones. It rarely is since almost all recordings are mixed for speakers. The original sound should not be the mental "starting point", but one option of listening competing against crossfed versions.

Try and error works. Recordings with consistent "spatiality" are easy, but some recordings have contradictory spatial signature and finding the optimal level of crossfeed can be a challenge.

The only kind of crossfeeder, that seems to work with "all recordings" using fixed crossfeed level ( -3 dB or so ) is "widefeeder", variation of Linkwitz-Cmoy with ~640 µs ITD. This is because for sounds coming from side, -3 dB ILD at low frequencies is enough * so you don't need stronger crossfeed for anything and large ITD is enough to keep the sound image wide for recordings which don't have that much excessive stereo separation to begin with. Room + speakers is also an ILD "regulator" which gives you almost the same amount of ILD (at least below 1 kHz) no matter what kind of recording you are playing.


* If ITD is smaller, such as typical 250 µs, stronger crossfeed (up to -1 dB) may be needed, because our hearing expects smaller ILD on sounds coming from ahead rather than form sides.[/HR]

 

[bigshot]

I don't know much about headphones and even less about cross feed, but I know there's a DSP that rechannels stereo to 5.1 for my speaker system that works with any stereo music. (Not so great for movies if you want the dialogue to be focused behind the screen.) So I would think that a one size fits all DSP for headphone listening should be possible. It would probably involve more than just cross feed though.

 

[71 dB]

I don't know much about headphones

Well, considering that you have made over 16 thousand posts on this board dedicated for the most part for headphones I might say you are either extremely modest about your knowledge or a really slow learner.

and even less about cross feed, but I know there's a DSP that rechannels stereo to 5.1 for my speaker system that works with any stereo music. (Not so great for movies if you want the dialogue to be focused behind the screen.) So I would think that a one size fits all DSP for headphone listening should be possible. It would probably involve more than just cross feed though.

Speakers are a bit easier in this sense because room acoustics will always polish out the most problematic spatial aspects of the sound. With headphones you need to get it right, because the eardums are an inch or two away!

 

[bigshot]

I used to be interested in headphones, but since I moved out of an apartment into my own home, I haven't had a lot of interest in them. I got a set I like and I use them with my computer. That is about it.

 

[WoodyLuvr]

Sacrilege! Burn the witch! Burn 'em all!

 

[castleofargh]

a real house without direct neighbors to try and kill you when you use speakers for too long? lucky!!!!!

Wade Wilson: I watched my own birthday party through the keyhole of a locked closet, which also happened to be my...

Vanessa Carlysle: Your bedroom. Lucky. I slept in a dishwasher box.

Wade Wilson: [Gasps] You had a dishwasher....

a one fit all solution would still need to have enough customization to fit all people and most headphone signatures. the delay alone should in the most basic concept of Xfeed, be based on sound sources at 30°angle on each sides(to match stereo speaker position), and how long the left speaker sound takes to reach the right ear after reaching the left ear. meaning it's directly related to the size of the listener's head.
and the same thinking goes for ILD cues. so we will always need some matter of customization and it's at least one reason why some Xfeed plugins happen to please some people and not others.

 

[ironmine]

Well, considering that you have made over 16 thousand posts on this board dedicated for the most part for headphones

Average 3.4 messages per day, for more than 13 years.
Wow. I am impressed. I would call it a full time job!

 

[bigshot]

Plus I was banned for over a year for pointing out that an emperor's nakedness! My stats slipped during the time I was on the bench.

 

[ironmine]

Does anybody use a reverb plugin before/after a crossfeed plugin?
Do we need to use any?
If you use one, which one it is and what settings you prefer?

 

[castleofargh]

I spent some time fooling around with the free version of reverberate. and I preferred to have it before the xfeed(after felt more natural for the reverb itself but not for the Xfeed). and last year I used the app that came with the waves NX head tracker, it had some some reverb along with a bunch of other things and I enjoyed it for a few months, then didn't(dunno why). nowadays I don't use any on the PC but I might come back to it anytime.
I still have some in Viper4android on my phone.

 

[Alexey Lukin]

So, why did you say that you know the RX De-Clip upsamples? It does not.

RX De-clip's post-limiter does upsample the signal, but only in the side-chain, for level detection. The main signal path is kept at the original sampling rate, both for clipping restoration and post-limiting.

The ITU defines a TP measurement as at least a x4 oversampling process based on a 48kHz sample rate, although Izotope's own dedicated TP meter/loudness plugin oversamples x9, it's entirely likely that it's De-Clip plugin does too.

This is correct: iZotope's true peak detection is usually based on 9× oversampling (it may depend on the product). One property of ITU's recommended 4× oversampling filter is that it has a ±0.1 dB passband ripple, which may slightly over- and underestimate “real-world” peak levels, esp. near the Nyquist rate (see the pic below). iZotope's upsampler has a flat passband, so its true peak indications may slightly differ from other true peak meters, while still being BS.1770 compliant.

BS.1770 true peak detection filter
​

 

[ironmine]

RX De-clip's post-limiter does upsample the signal, but only in the side-chain, for level detection. The main signal path is kept at the original sampling rate, both for clipping restoration and post-limiting.

Hi Alexey!

It's great to see you here!

 

[jgazal]

Great explanation. It gets even more interesting when you consider the Hafler PRIR. Basically you hear the sum from the center-front speaker (L+R to both ears), and you hear the differences from the center-back speaker (L-R to left ear, and R-L to right ear).

I made a Hafler PRIR where the center-back speaker was actually measured in front, but I turned my head in the opposite direction. I looked right instead of left and looked left instead of right. This way, the center-back speaker has the same spectral balance as the center-front speaker, and head-tracking helps me distinguish which sounds are from the front versus the back. Maybe Smyth can add a Hafler mode that works for any PRIR that has a center speaker or a closely-spaced pair.

Some live recordings sound great with crowd noise and hall reverb that come from the back. Recordings that were matrix-encoded sound great as well, and you might not realize which ones until you listen to them this way.

In addition, for 4.0 or 5.1 recordings, the effect can be flipped for the two discrete surround channels, Ls and Rs. Basically you hear the sum from the center-back speaker (Ls+Rs), and you hear the differences from the center-front speaker (Ls-Rs to left ear, and Rs-Ls to right ear).

Erik, please excuse me to bother you again.

Have your tried to measure an one real speaker crosstalk free PRIR with four virtual speakers (Front-Left and Front-Right at 0 degrees and Back-Left and Back-Right at 180 degrees azimuth), using the same approach you described, i.e., Back-Left and Back-Right virtual speakers are actually measured in front, but with your head turned in the opposite direction? Ipsolateral speakers at full mix.

On a second thought, maybe just one left speakers at -90 degrees and right speakers at +90 degrees?

Do you believe binaural or (even stereo recordings) may benefit from such arrangements?

 

[Erik Garci]

Have your tried to measure an one real speaker crosstalk free PRIR with four virtual speakers (Front-Left and Front-Right at 0 degrees and Back-Left and Back-Right at 180 degrees azimuth), using the same approach you described, i.e., Back-Left and Back-Right virtual speakers are actually measured in front, but with your head turned in the opposite direction? Ipsolateral speakers at full mix.

I have measured it two ways. One PRIR was measured with the rear speaker behind (180 degrees) and normal head-tracking. Another PRIR was measured with the rear speaker in front (0 degrees) and reversed head-tracking.

On a second thought, maybe just one left speakers at -90 degrees and right speakers at +90 degrees?

Speakers at +/-90 degrees would not work well for head-tracking.

Do you believe binaural or (even stereo recordings) may benefit from such arrangements?

The Hafler method works well for 2-channel sources. I have not tried other methods and cannot think of any that would work better.

 

[Jon Sonne]

Average 3.4 messages per day, for more than 13 years.
Wow. I am impressed. I would call it a full time job!

castleofargh, bigshot and 71 dB, I follow you guys, and I really appreciate your contributions to the sound science forum! @ironmine - I've just installed 112dB redline monitor, and I am really enjoying it with my new HD800S. The xfeed solved the problem with electric guitars sounding diffuse on the HD800S, now they sound powerful and crisp! I use the 90 degrees soundstage width and -1 dB phantom center. The speaker distance is turned off. In addition, I've added an EQ after the xfeed with a -5 dB peak at 5500 Hz and a -1 dB from 10kHz to 20 kHz centered at 13kHz. This works perfectly to remove the coloration in the treble. Thanks for recommending this wonderful vst plugin!

Merry xfeedmas!