I'm looking at "Tone of Voice and Mind" by Norman Cook, a work on how
emotion is perceived in speaking voices and in music.

Some recordings and audio systems sound more "real" to me than others.
What is "realism" anyway?

The human brain is very sensitive to "anamolies" in sensory input, when
different features on the input don't match each other. In Cook,

"If I announce to you in a cheery, light-hearted, bouncy tone of voice
that 'My grandfater died last night,' you are likely to ask me if I
have lost my mind, if I have myself fallen ill and if I understand what
I am saying. The anomaly will be inescapable for you... You will
undoubtedly not go through a conscious process of pitch analysis to
reach your conclusion, but you will 'in your gut' that something is
amiis if the literal linguistic message is about death, and yet the
tone of voice converys a happy emotional state, or vice versa. When
words and intonation do not match, the fact that emotional messages can
be transmitted through pitch is obvious."

Another example in a book by Daniel Stern about infant development:
infants, from a very young age, can link up what they see and what they
hear. They show a preference for looking at a movie that has
syncronized sound, as opposed to a movie in which the sound and video
are off.

When an audio system X doesn't sound real to me, it feels like this: in
my gut I know that something is amiss, something doesn't match up to
life.

I'm sure we can agree on that. For instance, compare the experience of
hearing a musical instrument live in your room, and hearing a
loudspeaker play in your room. I think most of us would agree the live
instrument sounds more "real" and "present"; and I hypothesize this is
because the brain can detect the way the features of the loudspeaker
signal don't quite "line up" in the same way.

Notice that this "lining up" of features can happen widely across the
brain. Listening to music involves noticing features and relationships
in the music, as well as noticing body sensation and emotion. I think
it is a good assumption that greatest realism is achieved when these
features correlate properly across the whole brain.

I'm inventing a concept: correlation of global musical features, or
CGMF. What this refers to the correlation of features and
relationships in the signal that affect the whole brain (hence
"global"). An accurate audio system would be one that preserves CGMF.
An inaccurate system would be one that screws up the relationship of
the features, or screws up CGMF.

I think that the objectivist would reply with, "No, an accurate system
is one that measures better than we can hear." That's a convincing
statement, but let's agree that there are some complex concepts lurking
behind that. There's the question of what signals we can hear and how
that knowledge was obtained. There's the question of what measurements
we can perform and how they relate to CGMF.

The objectivist is implying, "We know all that stuff well enough that
we don't need to explicitly mention it any more." But let's agree that
the statement "an accurate system is one that measures better than we
can hear" is founded on the security of our knowledge as mentioned
above.

In a test in which the subject doesn't access their emotions or body
sensations, I wonder how much that can say about CGMF, since so much of
the experience is left out.

-Mike