19 Oct
2016
19 Oct
'16
11:19 p.m.
On Wed, 19 Oct 2016 16:43:18 -0400, Jeremy Carter wrote:
On Wed, Oct 19, 2016 at 4:40 PM, Fons Adriaensen
wrote:
If this kind of artificial intelligence would have all the required
skills, it would work towards an imagined final result. Human audio
engineers have different tastes, too, so I don't expect that AI needs
to satisfy my taste, but I expect that such a tool is able to finish a
mix in an educated manner.
If the tool just suggests something and claims this to be useful
starting points, without the ability to finish the mix on its own, then
the useful stating points are fraud. If you don't have the skills to
finish the mix, how would you know what would be good starting points?
It's ridiculous, because we are that far away from the required
knowledge about audio and artistic taste, we need to have to program
such a tool, with what ever kind of perfect artificial intelligence,
that already might be available.
It's like preparing a manned spaceflight to Alpha Centauri, while we
are still far away from having the skills to do manned spaceflights to
Mars.
This has nothing to do with being afraid that technology could put
audio engineers out of work.
Regards,
Ralf
On Wed, Oct 19, 2016 at 10:25:26PM +0200, Ralf
Mardorf wrote:
Of course audio engineers would think something which could
potentially put them out of work was stupid (or something they *think*
could never put them out of work). What does the AI community think of
it though? The audio engineer with the promotionally
effective name demands
another audio engineer for this kind of assistance. IOW a human
does the mix and another human does the fine tuning. This is quasi
the counterpart of what Fons already mentioned. Let iZotope's Track
Assitant/Neutron do the mix, then use iZotope's Track
Assitant/Neutron to do the fine tuning of the mix it produced
before.
I spent most the evening with a group a sound engineers, discussing
things ranging from mixing desk ergonomics to how to approach a
particular mix that our host presented as an example of his work.
Inspired by this thread, I dared to coin the question 'how would
you feel about an EQ plugin that analyses the track it's in, gets
information from its peers in other tracks, and then adjusts
itself ?'
Result: general hilarity. Consensus was that this the most stupid
thing ever suggested. Had to take some action to avoid damage to
my own reputation.