Originally Posted by TheWuss
Saying that it "blends" the left and right channels is a bit of a misnomer. and thus one of the reasons crossfeed is misunderstood as a "mono-izer".
Audibly, the left and right signals are not blended, but are actually pushed even further apart. Because they are brought into focus, and the ear can pinpoint an acousstic location for the signal. In other words, what once sounded as if it was eminating from within the headphone now sounds as if it is eminating from further away.
Crossfeed actually does mix the channels together but it doesn't do it indiscriminately. The level of crosstalk varies with the frequency and drops off as the frequency increases. This mimics the way sound reaches you ears IRL. Lower frequencies with longer wavelengths will bend more around the edges of obstacles, such as a person's face. This means that the lower the frequency the more you will hear it in both ears even if source is at an angle to you and one of your ears doesn't have a straight line path to the source. These difference in volume between each ear that vary with frequency and angle are one way that your brain determines what angle a sound is coming from.
This centers the imaging and moves the soundstage forward because stereo mixes assume acoustic crosstalk from speakers which headphones lack. A guitar panned most or all the way to one channel will still be heard with both ears when played over speakers from diffraction (the sound bending around corners) even if you're in an anechoic chamber. With headphones you either won't hear it at all in one ear or the volume in the other ear will be lower than how you would hear it IRL. A hard pan won't occur in nature but even something panned most of the way to one side will sound like it's at a greater angle on headphones than speakers because the level in the opposite ear will be lower. Increasing the crosstalk between channels as frequency decreases helps to mitigate this by making the differences more natural and moving the stage forward.
It's not perfect though. The usual crossfeed circuits you see on most head amps aren't the most accurate simulation of what the sound waves do IRL. They capture the broad brushstrokes but don't add the fine details. There are other things that your brain uses to localize sound as well so even a perfect crossfeed won't give you perfect localization.
Originally Posted by edstrelow
The recording sounds interesting. Your psycho-acoustic explanation is off the rails. Bianuaral recordings do not provide crossfeed, nor does regular stereo, listened through headphones.. The sound of each channel heard by the opposite ear is an artifact of speaker reproduction often called a "phantom channel." Far from enhancing locaization it messes it up as anyone who has heard speakers that suppress it knows. Check out Polk's old SDA systems. Blending and mixing up the 2 channels simply makes the minimizes the stereo localization cues. The headroom people were simply full of beans on this issue and used this erroneous bit of psuedo science to peddle their cross-feed blend systems. If you like blend and crossfeed that's your choice, it will shrink the width of a stereo image and some people seem to be able to convince themselves that this give better forward projection of sound. In tuth it does sound more like speakers but speakers give flawed stereo.
I think you're probably the one full of beans on this issue.
Stereo music is already mixed with the acoustic crosstalk from the opposite speaker in mind and accounted for as much as possible. It's not perfect but it works pretty darn well in a well set up listening room. Crossfeed is certainly not as good as a proper binaural recording which encodes a more accurate transfer function than a basic modified Linkwitz filter can. Unfortunately pretty much all 2 channel music is mixed in stereo so we have to make due with what actually exists.
The reason to add crossfeed is simple. You hear sounds IRL with both ears even when they come from a single point. Excluding the very softest sounds and shortest distances there is no such thing as a hard pan IRL like there is when listening to stereo recordings over headphones. Reflection, diffraction, and even the attenuated and delayed wave which will pass though your head ensures that you hear the majority of sounds with both ears, especially loud sounds. Headphones just can't do that when fed stereophonic material. They sound loud when they only have to move a tiny bit of air right next to your ear but they don't put out nearly enough energy for the waves to audible after they get to the other side of your head. This asymmetry that headphones create with stereo recordings is completely unnatural. Stereo mixes that sound perfectly natural on speakers will commonly create soundfields that are physically impossible when listened to over headphones. A fly can buzz in your ear, sound loud by proximity, and cause an annoying asymmetry but it still sounds like it's right next to your ear. There is no instrument or voice which can be heard loudly in one ear, sound many feet distant, and simultaneously be inaudible in your other. Congratulations if you're able to get used to that. I get headaches from much more subtle imbalances. It certainly opens up your options when shopping for amps.
Crossfeed is indeed a compromise. A few simple LCR filters don't accurately reproduce even a generic HRTF particularly well. Even DSPs like TB Isone aren't perfect. Some people might ask why we should bother if the only solution is such a compromise. They probably haven't noticed that listening to stereo music over headphones is already a compromise, a far larger one than the "phantom" channel of stereo speakers IMO. Ultimately you've got to pick your poison. If you're not bothered by unnatural interaural level differences and prefer a wide soundstage over a deeper one then crossfeed isn't for you. If you are bothered by those you can use crossfeed and trade some soundstage width for it. Even binaural isn't perfect as individual HRTFs can vary quite a bit from the averages used to construct HATS. The fewer compromises you make the less music you'll have left to listen to.
You can talk about it's limitations or just not like the effect but it has a very firm basis in psychoacoustics.