[ExI] AI poker

[William Flynn Wallace]

A trained ear can transcribe either a sax solo or a French horn solo.  Put
them together, the human can somehow tell which instrument is which from
its unique timbre and work out who played what (not on the first try, or
even the second.)  A machine is nowhere close to being able to do that I am
told.  Or if so, it is a recent and very impressive development.



spike


I still don't see the problem.  For one, a sax has, it has been described,
a very muddy set of overtones (clearest is the oboe), and a French horn
(which is really English, whereas an English horn is really French - go
figure) has a much clearer set, so it should be easy for the AI to tell
which is playing which note.  Put two saxes together and it's impossible -
or is it?  Each horn has a slightly different set of overtones - no two
instruments are the same, eh?  I'll bet an AI could distinguish two
instruments that a person could not.


Mozart did something like this, only with a full orchestra and chorus.  The
Pope (or some really big dude) had a piece of music that was only played
for him, and Mozart attended a concert of it, and later wrote down every
note by every player or singer.  Now that's impossible even for most genius
musicians.


Most composers would not write for the sax because of the muddy overtones.
Not a quality instrument, they said.  Probably the reason people like it
(though, sadly, I don't except for Paul Desmond - Boots Randolph is a
nightmare to me).  I wonder why the oboe is not a pop instrument?


bill w

On Wed, Jan 25, 2017 at 2:26 PM, spike <spike66 at att.net> wrote:

>
>
>
>
> *From:* extropy-chat [mailto:extropy-chat-bounces at lists.extropy.org] *On
> Behalf Of *William Flynn Wallace
> *Sent:* Wednesday, January 25, 2017 12:13 PM
> *To:* ExI chat list <extropy-chat at lists.extropy.org>
> *Subject:* Re: [ExI] AI poker
>
>
>
>
>
>
>
>
>
>
>
> I am confident no one has been able to write code to read tells because we
> humans don’t know how we humans read tells.
>
>
>
> Another example of it is how a human can listen to a French horn and a
> saxophone play a duet, then write out the score, even getting the key
> signature right.  Not on the first try of course, but I can do that if I
> can play it several times.  With all our sophisticated Fourier analysis and
> all our computer whiz-bangery, we are nowhere close to even that.  Reason:
> we don’t yet know how our minds can do that task.  But I can assure you, it
> can do it, and I can demonstrate it.
>
>
>
> spike
>
>
>
> ​>…Huh?  Whataya mean we don't know how players read tells?  I'll bet if
> you ask them they can explain just what Joe did that they read as a bluff
> and they were right.  Are you saying they just have vague feelings but
> don't know what they are basing them on?
>
>
>
>
>
> The poker player can describe what Joe did, but there is no known way to
> turn that observation into code.
>
>
>
>
>
> >…Also - what's the big deal about music transcription​?  I'll even bet
> that some of the really big music composition programs can be fed music and
> transcribe it into the score…
>
>
>
>
>
>
>
> Indeed?
>
>
>
>
>
>
>
> (a quick google search) -
> The Best Music Notation Software of 2017 | Top Ten Reviews
> <http://www.toptenreviews.com/software/home/best-music-notation-software/>
>
> www.toptenreviews.com/*software*/home/best-*music*-*notation*-*software*/
>
> 1.
>
> You can plug your MIDI controller into your *computer*, and as you play
> notes, the *software* will add notes to the *sheet music*, creating *sheet
> music* as you play.
>
>
>
> ​Or is this something different from what you are talking about?
>
>
>
> bill w​
>
>
>
>
>
> Ja, different.  If you have a MIDI keyboard, the software can record what
> you played, or if you have a MIDI-equipped instrument such as an electric
> saxophone, the software can record what you played.  But that isn’t done by
> listening, it records key motion (and air pressure if it is an electric
> sax) like your computer does with the keyboard.
>
>
>
> If you have a microphone attached to both instruments, software **might**
> be able to separate the two, or if you have a single instrument playing at
> a time, software can transcribe it.  But… a human can do what we still
> haven’t been able to do with software: listen to a French horn/sax duet a
> few times and figure out the score.
>
>
>
> I chose those two instruments because they are generally in the same
> octave and it sounds cool if you get two good players doing a FH/sax duet.
> Software might be able to extract the parts in a piccolo/Tuba duet (anyone
> know?)
>
>
>
> A trained ear can transcribe either a sax solo or a French horn solo.  Put
> them together, the human can somehow tell which instrument is which from
> its unique timbre and work out who played what (not on the first try, or
> even the second.)  A machine is nowhere close to being able to do that I am
> told.  Or if so, it is a recent and very impressive development.
>
>
>
> spike
>
>
>
>
>
> _______________________________________________
> extropy-chat mailing list
> extropy-chat at lists.extropy.org
> http://lists.extropy.org/mailman/listinfo.cgi/extropy-chat
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.extropy.org/pipermail/extropy-chat/attachments/20170125/5b6b6521/attachment.html>