Subjective listening test results comparing conventional 2 speaker stereo and 3 speaker stereo with matrix.
These listening tests have two focus areas: Perception of comb filtering generated by stereo field interference and perception of phantom images.
The frequency band in these test is limited to high frequencies. This was decided due to known problems existed in high frequencies in the conventional 2 speaker stereo triangle.
Also the purpose of these tests is to find out if 3 speaker stereo with matrix can improve the problematic situations of 2 speaker stereo.
The typical problems in conventional 2 speaker stereo at high frequencies include the following:
The figure shows the loudspeaker configurations used in these tests. Also the loudspeaker signals are shown, and for the 3 speaker stereo they are generated by the matrix x=0.5.
The room is a typical living room with size of about 25 m2.
There are three cases:
1) The head is oriented forwards and laterally shifted.
2) The head is oriented sideways and laterally shifted.
3) The head is rotated but held in the same location.
The lateral shift is moderate, less than 0.5 m in either side.
These results define in which degree of severity the comb filtering is perceived in given circumstances.
The table below summarises the results with mono band limited pink noise.
The frequency band is divided in octave bands except for the top band which is a bit wider.
Listener discomfort rating scale is used for ranking.
The table includes the results for 2 and 3 speaker stereo confugurations for the three head orientations.
This is the same test as above but now the test signal is band limited mono music.
These results define the perception of phantom imaging when band limited stereo music signal is used.
The frequency band in this case is limited between 1 - 20 kHz.
The table below highlights the main perceptual differences between the 2 and 3 speaker stereo configurations.
The perception of comb filtering with mono band limited pink noise is considerably reduced with 3 speaker stereo with matrix than what is perceived with conventional 2 speaker stereo.
The perception of comb filtering with mono band limited music is, however, almost the same between 2 and 3 speaker stereo configurations. In both cases the comb filtering is not particularly strongly perceived.
The difference between the results obtained using pink noise and music is believed to be due to fundamentally different nature of common music compared to pink noise. Typically music is consisted of temporal structures whereas pink noise has more of a steady state characteristics. A valid question could be presented if pink noise should be used at all for audio tests. However pink noise is commonly used, and as seen quite remarkable differences can be found between 2 and 3 speaker stereos.
The phantom imaging test was interesting as the 3 speaker stereo with matrix is perceptually better in every regard than conventional 2 speaker stereo.
Overall, these tests would indicate that at high frequencies 3 speaker stereo with matrix is able to overcome some of the basic limitations of 2 speaker stereo.
It would propably not even be considered as an overstatement if one would recommend the 3 speaker stereo for home listening scenarios over the 2 speaker stereo.