I couldn't find Dolby Headphone wrapper for Linux, so I came out with some idea how to add virtual environment using JACK and LADSPA plugins.
Below is the screen-shot.
I connect output from Audacious to input of channel "1" on JackMixer, and then I connect output of that channel to JackRack where I set delay and filter. Then and I connect output from that JackRack to input of channel "2" on JackMixer swapping L with R. On the JackMixer I set volume of channel "2" to -1.5dB as I think this is the difference between how much you hear left speaker in your left ear and in your right ear. So this simple trick perfectly simulates speakers being in front of you.
Next trick is to simulate room in which you are listening. I want to simulate echoes and resonances from walls. So I assume sound first hits wall behind speakers and then it hits the wall behind the listener. I could complicate model much more, but I already find hard to manage Jack connections, so I stuck to this simplified model. In order to implement it I connect output from channel "1" on mixer to another JackRack, where I set several effects. I set delay to obtain echo, and I set low pass filter to simulate damping, and then I set comb filter to simulate resonance from walls. The output from that rack I connect to input of channel "3" on mixer. Next I want to simulate wall behind the listener. Again to do that I take output from channel "3" on mixer and connect it to another JackRack, where I set delay and matrix spatialiser, which inverts the phase so that sound seems to be coming from behind the listener. The output goes to mixer channel "4".
I set all parameters empirically, so scientifically looking at them they can be completely wrong, but to my ears result is IMHO very cool, actually I think I might prefer it over Dolby Headphone.
Here is screen with settings I came up for rock genre: