In my experience the term DSP refers to Digital Signal Processing and is used by companies like Yamaha to describe processors and multi-amp set-ups used to synthesize ambience effects of concert halls, stadiums and the like using time delays and other proprietary means with multiple amps and speakers. I have had 2 such systems myself and quite like them. They also do Dolby very well.
I don't regard the role of headphones as trying to sound like speakers though. For a start, straight speaker stereo has serious issues of artifacts due to the phantom channels whereby the left ear gets both left and right channel signals and similarly the right ear gets both channels in which the opposite ear signals is time delayed due to the extra time the phantom channels take to reach the opposite ear, due to the extra dstance these signals travel to reach the opposite ears. Speaker stereo is inherently muddled, which you will realize better if you have heard any of the systems, such as the old Polk SDA speakers, which attempt to supress the phantom channels. I have a set of these in my main speaker set-up and they are quite effective, including having the ability to project sound outside of the actual speaker positions. But what the Polks end up sounding like is more like headphones, since they are essentially trying to present an uncompromised headphone type of stereo signal in which the left and right ears only get pure left and right signals.
The fact that speaker stereo works at all has always struck me as a testament to the ability of our brains to resolve ambiguous auditory information. But just about any time you take an auditory signal and make two channels of it slightly different in characteristics and play these back by 2 drivers, speakers or heapdhones, you will get some sense of stereo. One of the early tricks to turn monaural sound into pseudo stereo simply involved taking a multiband 2-channel graphic equalizer and setting the bands differently for the 2 ears and voila it sounds like stereo.
The advantage that even ordinary speakers have over headphones is that they provide externalization of sounds because speakers are located well away from the ears. Speaker sound is not bound up in the listener's head as happens with headphone listening.
As far as cross-feed or blend is concerned, as I note above, what you really get in speaker listening is not just cross-feed but discrete phantom channels, with time delays. Why anyone would want to simulate this psychoacoustic mess with headphones beats me.
What I think people are really hoping to achieve with crossfeed is some sense of externalization of sound but I don't see that as te way to go. I know that when I had a blend system I almost never used it because it added nothing pleasurable to the listening experience. All it sounded like to me was weaker stereo imaging and if you mix the left and right channels completely you will obviously end up with monaural sound which would be completely pointless. But if crossfeed sounds better to you then so be it.
Binaural sound is a lot of fun. It can give an uncanny sense of realism through headphones but even then I do not find that it projects sound well to the front. I remember an old Sennheiser display disc in which they acknowledged a tendency to project sound more to the rear than to the front even with binaural recordings.
As far as giving a sense of externalization to headphone listening, the Sigma and Akg K000 phones position the drivers ahead of the ears and to give you a flow of sound past the ears similar to sound sources in front of you. They will to some extent give you a head-related-transfer-function http://en.wikipedia.org/wiki/Head-related_transfer_function comparable to a sound source in front of you.
You might try to simulate such HRTF effects with digital processing however since these are to a large extent unique to each person, I would think you are only going to get part way. The Sigma and AKG at least allow each listener to simulate their own HRTF with their own heads.
