Originally Posted by jj_0001
The elements of the performance that they are listening to are well, far above threshold of audibility, and they can hear what they need to hear, modulo ideas about particular instruments, without a good system.
Things like pace, rhythm, timing are characteristics that are well above threshold and can conveyed with 100-7kHz bandwidth and not the cleanest of signals.
The worst speaker in the world does not cause the violin player to sound like he or she has bad intonation, just maybe a bad violin, for instance.
Another way of looking at my proposition is that the musician is listening to himself through the reproduction mechanism. Let us assume a musician of the highest musicianship and highest integrity (a.k.a. he's a snob). He records a cut and while his performance is still fresh in his head he listens to the recording of himself which has been processed through the reproduction chain all the way through the speakers to his ears. I have intentionally frequency shifted up two notes worth. When I ask" is this what you played" he will say no. the question "is this the same as what you played?" is pretty loaded. relative to me when he listens to the playback he is listening for elements beyond my capacity. If it is his intention to communicate those elements and they are somehow "lost" and he is of the highest integrity he will say no. If he doesn't have the greatest in confidence in his memory he will want to re-record the passage on the assumption the loss of transmission was from his head to his trumpet.
when I was in highschool I went to a jazz festival and the headliner was Bill Watrous. there was a buzzing in the speakers and he kept glaring at the sound technicians to fix it. The reasons why he was angry about it only be speculated. Why would he care about the fidelity of the amplification?
My question was deliberately open to interpretation because I didn't want to bias what information content the listener was trying to detect. It is Yes or No answered because a binary response relative to me is in very similiar units. The question is simple enough so that all potential consumers can undergo the same testing procedure and understand the question as the musician. So when a consumer undergoes the evaluation and answers Yes and the musician answers yes, the units of the final transmission are the same. anything lost in the consumer is not the fault of the reproduction chain but lost from the consumer. However, the transmission from musician to consumer relative to me
Where the psychoanalyst comes into play is that the test procedure has to have very carefully devised questions. Answering the questions the musican will be thinking of self to self, self to consumer to self, self to technician to self..etc. The consumer too has to understand the questions to result in the same units as the musician otherwise the test is inherently flawed.
(An interesting consequence of this is that relative to me I cannot create a 100% transmission mechanism from an english musician to a japanese consumer, but relative to an english/japanese translator the transmission mechanism can be 100%)
Where the Neural network scientist comes into play is that their expertise is in creating weighted values based on binary input. They also can create optimized search algorithms. If I'm really looking to quantize the importance of phase relative to magnitude (relative to the consumer, relative to me) then a fuzzy logic algorithm can do it.
Another consequence is that to ensure 100% transmission from musician to consumer (relative to me), at the conclusion of the tests they must both answer Yes to all of the questions and flow through the same test algorithm path. The inherent data of interest to the engineer are the weights required to force the question path to the same conclusion. Those are the minimum requirements of 100% fidelity from musican to consumer relative to the engineer.