- Joined
- Jun 19, 2018
- Messages
- 6,652
- Likes
- 9,406
- Thread Starter
- #141
I don't think it is flawed as that: stereo imaging from speakers doesn't float around dependent on the frequency content. You might try to make a case that the brain 'latches on' to some content and from then on places the source from that point, but I don't think it is as fragile as that - you will always hear it from the same place wherever in the recording you start.
That's true so long as your brain perceives the combination of frequencies as a unified "voice". But where the content of a "voice" is primarily in the lower frequencies and does not contain a transient (e.g. low voices, low bowed strings, bass synthesis etc.), our localisation of them may not follow the same principles as with a transient that contains high frequency content - or at least, there's good reason to believe it might not.
The thing about Blumlein-style stereo (e.g. pan pot) over speakers - which I intend to investigate further - is that the crosstalk combined with the direct sound for each ear combines to give an actual, physical time-of-arrival difference that is far stronger in effect than a vague volume difference....
It is not saying that a volume difference is interpreted as a direction, but that the volume difference creates an actual, physical time-of-arrival difference at the listener's ears. The icing on the cake would be if that difference was more-or-less stable with head movement and/or the listener getting up and walking around - in the right way i.e. the source seeming to stay at a fixed place in space even when the listener turns their head.
This is correct, but the same effect is present in any recording method involving stereo, and is more pronounced in the methods other than Blumlein which add an additional phase difference between the two channels (i.e. additional to that created by the path length difference between the listeners' ears and each speaker, which is all that Blumlein relies on insofar as phase is concerned).
So I mostly agree with your arguments, but still believe that there are aspects of stereo localisation that may be enabled better by other recording methods, and that this experiment did not investigate these adequately to draw conclusions (apart from the in the case of transients containing HF content, which accounts for most percussion, and can probably be extrapolated to instruments that consist of a strong attack followed by sustain and/or long decay given that transients - when present - tend to dominate in terms of our localisation of a sound).
The authors inadvertently acknowledge this limitation by stating in their beginning assumptions that transients predominate in localisation (so what of sounds - human voices, bowed strings etc - that do not contain transients?) and by limiting their experiment to transients on the basis of this assumption.
Last edited: