| Richard Wentk commented on my request for information on using sones to set
amplitude in csound:
>Hey, I'm possibly missing something here, but isn't this a little bit of a
>wild goose chase?
was wondering what it was going to take to stir this nest (wink).
>Subjective loudness doesn't just depend on amplitude and harmonic content
>but also on musical context. Instruments in ensemble appear to mask each
>other, which is why engineers spend so much time tweaking EQ and level, and
>conductors spend so much time getting the strings (etc) to play louder or
>quieter.
>
>I'd be very surprised if it proved possible to concoct a nice simple
>equation that described how your brain responds to this process. Modelling
>sine waves is a waste of time because very little music uses pure sine
>waves,
It may be that i'm missing something. currently i am experimenting with
setting levels of sine waves using GEN 10. just keeping it simple, but do
understand that a bit of chorus or vibrato can be useful. trying to build
my perception. is my impression of an advantage in the use of sones vs
decibels somehow completely nullified at the moment a single overtone is
added? or two? or seven? that fletcher-munson curve sure did scare the hell
out of me.
>and once you start adding time-varying partials and these
>psychoacoustic masking effects the picture gets too complex to model
>effectively. (Does the attack have more of an effect than the sustained
>portion of a sound? What about inharmonic sounds? And so on...) Then there
>are also questions about phase, about phase-correlation, and absolute
>pitch. (E.g. for audiences used to equal-tempered music, playing slightly
>sharp seems to be a good way to get extra attention. And of course there's
>also compression, which can make things sound louder without changing the
>peak level.)
>
>And here's the killer - everyone's hearing varies anyway. There really
>isn't any such thing as an 'objective' musical experience. One person's
>'too loud!' is another's 'about right.' (How else can anyone explain heavy
>metal? :) )
>
>There are even differences in *how* people hear and what they listen out
>for. Some have a superb sense of rhythm but no harmonic sense. And vice
>versa. Some - poor souls - have neither.
this is a question: is the above getting excessively skeptical for my
purposes? the reason for my suspicion is that, ten years ago, i was
surrounded by an ornery band of audio aficionado / technicians who insisted
that quad audio was a waste of time in a concert hall because each member
of the audience would not experience it in exactly the same manner.
>So isn't it more practical just to stick with some approximate level
>indicator like ampdb and then adjust by trial and error till it sounds
>about right? The problem with a rigorous mathematical treatment is that it
>might make it too time consuming to actually get any music done. :)
ask me this again in a few years. just using decibels now. currently on
page 12 of five 1000 page books as i conduct my experiments. and i have a
day job completely unrelated. my guess is that there is a healthy mix of
subscribers present. just plain musicians, computer programmers, hardware
developers, mathematicians and others with equally useful skills in this
arena. and a few who have it all together.
all of the above are capable of making art. as a just plain musician
obsessed with the control and realization of musical visions, i find myself
on a point on a path lacking a clear view of how i will make art with this
thing sitting on my desk. pure chance of course is not my bag. maybe the
automatism of the surrealists (chance/subconscious followed by control).
but probably calculation of little things followed by chance/subconsious
followed by control. and this brings us back to sines sones sines
sones.
but now let's back up to those computer programmers. as it is clear that we
must choose some scale to measure volume, why not choose the most
intuitive? can't we just assign all that rigorous mathematical treatment to
the computer? perhaps not a priority for csounders at the moment. but i
*know* that there are developers of commercial music applications
monitoring this list for ideas. welcome. if i'm not completely soak and wet
on this one perhaps they may experiment. if they haven't already
implemented this idea years ago. and if i am drenched, perhaps the
marketing departments of their respective firms will get wind of this and
force them to implement SoneSational Amplimetric Parameter Settings anyway.
sorry.
oh yeah. and while you're at it...
definition of crescendo: to get louder musically. an exceptionally skilled
tailor once showed me the secret of removing enormous amounts of material
from oversized clothing without causing the fabric to pucker: a gentle
curve drawn into a straight line. all freehand from point to point. but
sometimes puckering is desirable. of course this choice does exist in some
fashion in csound as well as in applications with graphic interfaces. just
some words of desire. if there has to be a default, extend best efforts to
make it a good one. and then add an override as soon as possible. if there
doesn't have to be a default, create an optional one anyway. the holy
grail: total power to content providers for self-selecting time-sinks.
reporting for jury duty next week. maybe i'll get it all sorted out over
the next sensational murder trial, write a book, appear on talk shows and
thereafter properly dedicate my life.
all the best,
tolve
|