Skip to main content
Topic: Crossfeed - The next generation (Read 12465 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

Crossfeed - The next generation

After much experimentation I arrived at something that will expand the crossfeed concept but isn't as radical as Dolby HP. Instead, I use the BS2B with *Foo_channel mixer. The aim is to not only cut hard panned sounds but to simulate a listening room setup with superior imaging, unlike the spacious but broken presentation of headphones. Many audiophiles will argue otherwise, But I personally know this is possible from experimenting with dolby hp. It is already a reality in Samsung TV's - called SRS surround.

The idea is :

1) Cut the super-stereo effect with Bs2b , Otherwise additional DSP's will sound distorted or artificial and hard panned sounds will be harsh.

2) Simulate a listening room. Using *Foo_channel mixer the stereo-width is expanded (up to 1.65 for my personal taste). Taking the concept further: 2ch stereo is upmixed to 4ch to provide a better 3d-spatial effect.

3) Many headphones do not have a flat response. There are spikes in the mid treble around 7khz but often occur from 2~10khz. This results in an aggressive treble that will cause fatigue and ruin imaging. If this is the case one should apply EQ to the HF region to get a smoother treble. I have had to do it to the Alessandro MS-1 and to a lesser extent to the Sens PX-100 .


After all is done the foobar DSP setup is:

- **Bs2b (c-moy or JM) > channel mixer (4ch or more, stereo width = 1.65, upmix=copy or surround)

** must be the 1st plugin in chain if using > 2ch

* link: http://wiki.hydrogenaud.io/index.php?title..._channel_mixer)



I welcome feedback.
wavpack 4.8 -b3hx4c

Crossfeed - The next generation

Reply #1
Sorry for being direct, but I think that sounds horrible.

The increased stereo mix will just boost the side signal (L-R) or cut the mid signal (L+R). Why upmix to 4 channels? How are they downmixed onto your 2 headphone channel, if at all (in which case surround instead of copy would just further distort the sound)?


Have you tried a reverb plugin after the crossfeed and EQ after that?
"I hear it when I see it."

Crossfeed - The next generation

Reply #2
Sorry for being direct, but I think that sounds horrible.

The increased stereo mix will just boost the side signal (L-R) or cut the mid signal (L+R). Why upmix to 4 channels? How are they downmixed onto your 2 headphone channel, if at all (in which case surround instead of copy would just further distort the sound)?


Have you tried a reverb plugin after the crossfeed and EQ after that?



Yes I know it sounds bad. But that is where the crossfeed comes in - to cut the bad effect.  I haven't tried reverb yet. Its almost unreal vs switched off.  the 4ch upmix sound more 'full' or '3d' to my ears. Again the effect is subtle. Switching to 2ch + bs2b sounds more hollow but still much better than default headphone sound.
wavpack 4.8 -b3hx4c

Crossfeed - The next generation

Reply #3
I have switched back and forth many times and am astonished at the great result. The center is there unless the stereo-width is 2.00 or close. The result effect is similar to my Samsung TV .

If you think about it its like improving subjective quality even at the expense of objective quality. Just like psychoacoustic.
wavpack 4.8 -b3hx4c

Crossfeed - The next generation

Reply #4
I have used bs2b as you suggested followed by the channel mixer to boost the side channel (stereo width 1.65) and the result is a mess imo.
There stereo image is all over the place and the end result is more out of phase than even the original stereo track.
"I hear it when I see it."

Crossfeed - The next generation

Reply #5
I have used bs2b as you suggested followed by the channel mixer to boost the side channel (stereo width 1.65) and the result is a mess imo.
There stereo image is all over the place and the end result is more out of phase than even the original stereo track.



Thanks for testing. There could be several things:

- I still use foobar 1.03. Maybe the DSP behave differently like bs2b isn't working with channel mixer. This can be checked by removing bs2b from the chain. On my setup it has to be the first one in the chain. If its not enabled then the stereo field is totally off as you describe.

- Stereo-width value is too much for the given setup - Try a lower one like 1.05 ~ 1.30

- Headphone needs EQ in the mid treble. This is the worst and most complex issue I faced thus far. If treble has spikes and not smooth then the sound will have a rough bombastic character.  You have to lose a bit of detail to  gain a better stereo image.
wavpack 4.8 -b3hx4c

Crossfeed - The next generation

Reply #6
After much experimentation I arrived at something that will expand the crossfeed concept but isn't as radical as Dolby HP. Instead, I use the BS2B with *Foo_channel mixer. The aim is to not only cut hard panned sounds but to simulate a listening room setup with superior imaging, unlike the spacious but broken presentation of headphones. Many audiophiles will argue otherwise, But I personally know this is possible from experimenting with dolby hp. It is already a reality in Samsung TV's - called SRS surround.

The idea is :

1) Cut the super-stereo effect with Bs2b , Otherwise additional DSP's will sound distorted or artificial and hard panned sounds will be harsh.

2) Simulate a listening room. Using *Foo_channel mixer the stereo-width is expanded (up to 1.65 for my personal taste). Taking the concept further: 2ch stereo is upmixed to 4ch to provide a better 3d-spatial effect.

3) Many headphones do not have a flat response. There are spikes in the mid treble around 7khz but often occur from 2~10khz. This results in an aggressive treble that will cause fatigue and ruin imaging. If this is the case one should apply EQ to the HF region to get a smoother treble. I have had to do it to the Alessandro MS-1 and to a lesser extent to the Sens PX-100 .


After all is done the foobar DSP setup is:

- **Bs2b (c-moy or JM) > channel mixer (4ch or more, stereo width = 1.65, upmix=copy or surround)

** must be the 1st plugin in chain if using > 2ch

* link: http://wiki.hydrogenaud.io/index.php?title..._channel_mixer)



I welcome feedback.


You're reducing the stereo separation in step 1. and then expanding it again in step 2, then up-mixing to 4-channels before down-mixing to your stereo headphones?  Did I read that right?

Crossfeed - The next generation

Reply #7
After much experimentation I arrived at something that will expand the crossfeed concept but isn't as radical as Dolby HP. Instead, I use the BS2B with *Foo_channel mixer. The aim is to not only cut hard panned sounds but to simulate a listening room setup with superior imaging, unlike the spacious but broken presentation of headphones. Many audiophiles will argue otherwise, But I personally know this is possible from experimenting with dolby hp. It is already a reality in Samsung TV's - called SRS surround.

The idea is :

1) Cut the super-stereo effect with Bs2b , Otherwise additional DSP's will sound distorted or artificial and hard panned sounds will be harsh.

2) Simulate a listening room. Using *Foo_channel mixer the stereo-width is expanded (up to 1.65 for my personal taste). Taking the concept further: 2ch stereo is upmixed to 4ch to provide a better 3d-spatial effect.

3) Many headphones do not have a flat response. There are spikes in the mid treble around 7khz but often occur from 2~10khz. This results in an aggressive treble that will cause fatigue and ruin imaging. If this is the case one should apply EQ to the HF region to get a smoother treble. I have had to do it to the Alessandro MS-1 and to a lesser extent to the Sens PX-100 .


After all is done the foobar DSP setup is:

- **Bs2b (c-moy or JM) > channel mixer (4ch or more, stereo width = 1.65, upmix=copy or surround)

** must be the 1st plugin in chain if using > 2ch

* link: http://wiki.hydrogenaud.io/index.php?title..._channel_mixer)



I welcome feedback.


You're reducing the stereo separation in step 1. and then expanding it again in step 2, then up-mixing to 4-channels before down-mixing to your stereo headphones?  Did I read that right?


Yes. The upmix is optional. Using 2 channel gives nearly the same result. The width value can be adjusted to suit ones personal taste. The result should be subtle : the advantages of crossfeed with a spacious sound.
wavpack 4.8 -b3hx4c

Crossfeed - The next generation

Reply #8
I have tested the setup with the current Foobar 1.3.8 and everything works identical to 1.03

I also suggest to let the ear / brain acclimate for several minutes with each setting as jumping from one to another often causes fatigue.
wavpack 4.8 -b3hx4c

Crossfeed - The next generation

Reply #9
Well, what's happening is this:

bs2b basically mixes a low-pass filtered signal into the other channel. With the additional delay it adds this means that you'll get the biggest reduction of stereo separation at lows and mids - approximately like we hear in nature.

Test signals is just an impulse in the left (red) channel. Without crossfeed we'd just see the left channel signal as a straight line. So far that's not so bad.


Now to the stereo width slider, that mostly boosts the side signal a lot. If we do that, we get part of the inverted left signal in the right channel, creating some kind of out of phase effect...


In the time domain:



This is known to smear the stereo image, but at least it's not as bad as an inverted channel here.
"I hear it when I see it."

Crossfeed - The next generation

Reply #10
The question is which is more real to the human brain / hearing system. Isn't the out of phase thing normal for listening room once the speakers become too apart ? Sure it all seems good on paper but there is still something missing for the brain to comprehend when using just crossfeed. It seems you have to break things to fix what is already broken by default then go with the best subjective result .
wavpack 4.8 -b3hx4c


Crossfeed - The next generation

Reply #12
The question is which is more real to the human brain / hearing system. Isn't the out of phase thing normal for listening room once the speakers become too apart ? Sure it all seems good on paper but there is still something missing for the brain to comprehend when using just crossfeed. It seems you have to break things to fix what is already broken by default then go with the best subjective result .


What kind of listening room are you trying to simulate?  If it's an "average" one, I don't see the point.  Are you talking about comb filtering from phase cancellation?

Crossfeed - The next generation

Reply #13
The question is which is more real to the human brain / hearing system. Isn't the out of phase thing normal for listening room once the speakers become too apart ? Sure it all seems good on paper but there is still something missing for the brain to comprehend when using just crossfeed. It seems you have to break things to fix what is already broken by default then go with the best subjective result .


What kind of listening room are you trying to simulate?  If it's an "average" one, I don't see the point.  Are you talking about comb filtering from phase cancellation?



A stock living room where you sit in front of speakers / home theater system. Why you don't see the point ?

The superstereo effect is wrong from a psycho acoustic pov. I may have created a 'superstereo' like effect but more correct for the human hearing system at least to my ears.

Samsung and other manufacturers seem to know this well and can get this into TV's , software dvd etc, works for movies but can also work for 2ch audio. The main hurdle is lack of interest / objection from audiophiles or HIFI enthusiasts  - lots will not accept any processing / EQ  or even a crossfeed.
wavpack 4.8 -b3hx4c

Crossfeed - The next generation

Reply #14
The question is which is more real to the human brain / hearing system. Isn't the out of phase thing normal for listening room once the speakers become too apart ? Sure it all seems good on paper but there is still something missing for the brain to comprehend when using just crossfeed. It seems you have to break things to fix what is already broken by default then go with the best subjective result .


What kind of listening room are you trying to simulate?  If it's an "average" one, I don't see the point.  Are you talking about comb filtering from phase cancellation?



A stock living room where you sit in front of speakers / home theater system. Why you don't see the point ?

The superstereo effect is wrong from a psycho acoustic pov. I may have created a 'superstereo' like effect but more correct for the human hearing system at least to my ears.


Unless you have properly-placed acoustic treatments or perhaps Linkwitz DIY speakers, an average domestic room is just not a good listening environment.

Crossfeed - The next generation

Reply #15
Sorry for being direct, but I think that sounds horrible.

The increased stereo mix will just boost the side signal (L-R) or cut the mid signal (L+R). Why upmix to 4 channels? How are they downmixed onto your 2 headphone channel, if at all (in which case surround instead of copy would just further distort the sound)?


Have you tried a reverb plugin after the crossfeed and EQ after that?


If he's trying to simulate a domestic-sized room, there won't actually be reverb.

Crossfeed - The next generation

Reply #16
It doesn't matter treatment or not. The point is to expand the existing bs2b soundstage to improve realism. My cheap Mazda car speakers will do over long term headphone sessions. It doesn't have broken super stereo sound [it does have other issues]. I can connect emotionally to the music much easier even listening to POS car speakers.

An average living is a not too bad enviroment IMO. I had one since 2001. I was just a matter of placing the speakers correctly and I could get fairly detailed sound without headphone quirks.
wavpack 4.8 -b3hx4c

Crossfeed - The next generation

Reply #17
Comb filtering or reflections maybe. I am not an expert by a long shot but some reflections can be very pleasurable or even essential to the hearing system. Of course too much of it blunts the image - this setup [or dolby HP / DTS CS] doesn't sound like that to me and plenty of speaker ones are more 'off'
wavpack 4.8 -b3hx4c

Crossfeed - The next generation

Reply #18
I know these algorithms, but they don't create atmosphere like that.

Here's a simple test file: pink-lr2.flac (1.92 MB)

The noise source should wander slowly and evenly from left to right from about 2.5 to 12.5s.
Try this without DSP, with crossfeed alone, and with increased "stereo width".

"I hear it when I see it."

Crossfeed - The next generation

Reply #19
Have you tried Razer Surround?  Maybe try up-mixing to 7.1 and run it through that.  Unfortunately, you will never simulate the bone conduction or chest cavity vibration with headphones.


Crossfeed - The next generation

Reply #21
Speaking off crossfeed, does anyone know of any software for crossfeeding all audio output under Linux or at least a crossfeed VLC-plugin for watching films?
Listen to the music, not the media.
Qualidade em MP3

Crossfeed - The next generation

Reply #22
I know these algorithms, but they don't create atmosphere like that.

Here's a simple test file: pink-lr2.flac (1.92 MB)

The noise source should wander slowly and evenly from left to right from about 2.5 to 12.5s.
Try this without DSP, with crossfeed alone, and with increased "stereo width".



I did. I am even more impressed than before. I don't see a major problem with my setup even if its a kludge.

a) Without crossfeed or any dsp the noise trails from L to C to R but the superstereo effect adds extra noise, panning or like a rumbling effect . So its NOT transparent and painful.

b) With crossfeed much more transparent and reduced pain.

c) With crossfeed + stereo width: Nearly transparent up to 1.20 with slightly added noise. At 1.65 its more noisy but this noise to me is more preferable than a).    At 2.00 it is very noisy and distorted.

d) Stereo Width without crossfeed: The problems of a) together with extra white noise. This is the most noisy and probably the worst.

I may add that the stereo-width adds white noise like character with I like more than an ugly rumbling effect of example a).

In fact example (a) shows the biggest bug with headphones: superstereo yet a MONO rumbling character that can be painful.
wavpack 4.8 -b3hx4c

Crossfeed - The next generation

Reply #23
I may add that speakers when properly placed are stereo without an added ugly mono character. Even when listening to near-mono stuff it sounds respectable. Headphones are more stereo in the sides (it can also hurt) yet exhibit an unpleasant mono center-channel illusion (vocals , bass).

Adding stereo width + crossfeed reduces hard panning (L R) AND improves the mono center effect to my ears. That = better spatial imaging.
wavpack 4.8 -b3hx4c

Crossfeed - The next generation

Reply #24
I'm not sure what you mean with unpleasant mono. Headphones are capable of real center-speaker like mono instead of the stereo phantom center.

a) There is no extra noise or rumbling. It's just that since there is no natural delayed crosstalk to the other ear, but instead silence, it sounds completely unnatural or "painful".

b) I wouldn't say it is more transparent, it's just more natural.

c) At 1.2 you can already notice a slight change in the panning. Instead slowly and consistently moving left-right it stays a bit longer at the sides and speeds up when moving from left-right.
At 1.65 you got a tonality change at both sides. The sound source stays way too long on each side (about 5 seconds) and in the the transition from left-right, which is too fast, it is a blurry mess.
Also, the first 4 seconds are basically like the first 2 seconds in a), except that there is a little sound in the other channel that gets rid of the painfulness.

So I don't see how c) is a positive change.
"I hear it when I see it."

 
SimplePortal 1.0.0 RC1 © 2008-2019