Thanks, interesting! Learning machines to compose might also be helpful for self-reflection what it is that one is looking forward to in music and what it's essence is.
It's so crazy what complexity is possible with the tiny amount of neurons that today's computers can handle :) It's like teaching a fruit fly! Very cool episode.
Fascinating! Matthew's machine reminds very much of a player mechanism that was situated over the piano keys, "playing the piano" with mechanical fingers (opposed to player mechanisms built into the piano that moved the action). The "learning process" of Tchaibotsky is essentially the same of a human composer: Starts out based on, or "inspired by" an existing composer, i.e. Scriabin with Chopin, and then going from there developing their own characteristic music. It would be really interesting to see Tchaibotsky's compositions in the long term!
Penetrating interview from Tiffany in this highly complex pursuit. Fascinating and skillfully made intelligible without becoming trivial with Tiffany not deriding this as an indulgent infringement of the high art of classical music. Fascinating thoughts on Artificial composition, the need for the highly intellectual, abstract and exacting engineering of synthesized sound is not required here with the sounds being 'mechanically' generated.
We're only about 50 years in the computers - in their infancy. 500 years from now I'll bet you computers do almost all the musical writing and playing.
The next step is making an ai where the robot could compose according to the composer mental state and experience(e.g. Schumann when he was fighting clara father vs Schumann when he was admitted to the asylum)
Music is related with a phenomena known as art, where we, as an emotional agent with complex general inteligence, can express ourselves through abstractions. This approach to music with AI makes the assumption that you can mathematically predict composer's feelings and musical intentions based on empirical musical data (scores and performances) and its relationships, using a powerful function aproximator called neural network. The problem is that these people are missing the most important data, which is the data that our brain uses to give meaning to music, and that composers use to compose, and for that you basically need to be able to reach human level intelligent robots.
Creative abstraction is something that is being explored but it is something innate actually and at present something only can be done in a limited fashion by AI implications
@@maxwellsim76 Thanks, nice answer. If it's innate then it's curious the complexity of Schoemberg and other state-of-the-art composers, it seems you need to do a lot of rational work to understand what they are trying to say there, or better said, to create your own version of what happens in their pieces. I keep thinking this complexity of art is only achivable by (at least) human level intelligence. Also, we do not want human level intelligent robots around, so leave art to humans please!!!
Tiffany, you are a great and intelligent interviewer. This was fascinating, but you seem to have the ability to make all of your interviews so.
Thanks, interesting! Learning machines to compose might also be helpful for self-reflection what it is that one is looking forward to in music and what it's essence is.
you know you have made it in life when you appear on classical chats
Hold up...
Very interesting chat Mr Graham!
I like "Tchaibotsky" and "Robotz-art" for a name.
"Ghost in the machine" ,very appropriate for Halloween, cool.
So amazing. Thank you so much Tiffany for your time, enthusiasm and kindness. Ji-Hae Park is also doing some AI creating. Thank you Matthew.
It's so crazy what complexity is possible with the tiny amount of neurons that today's computers can handle :) It's like teaching a fruit fly! Very cool episode.
And btw, congrats to 10.000 subscribers!
Fascinating! Matthew's machine reminds very much of a player mechanism that was situated over the piano keys, "playing the piano" with mechanical fingers (opposed to player mechanisms built into the piano that moved the action). The "learning process" of Tchaibotsky is essentially the same of a human composer: Starts out based on, or "inspired by" an existing composer, i.e. Scriabin with Chopin, and then going from there developing their own characteristic music. It would be really interesting to see Tchaibotsky's compositions in the long term!
Penetrating interview from Tiffany in this highly complex pursuit. Fascinating and skillfully made intelligible without becoming trivial with Tiffany not deriding this as an indulgent infringement of the high art of classical music. Fascinating thoughts on Artificial composition, the need for the highly intellectual, abstract and exacting engineering of synthesized sound is not required here with the sounds being 'mechanically' generated.
We're only about 50 years in the computers - in their infancy. 500 years from now I'll bet you computers do almost all the musical writing and playing.
The next step is making an ai where the robot could compose according to the composer mental state and experience(e.g. Schumann when he was fighting clara father vs Schumann when he was admitted to the asylum)
Very halloweenish I love it!
I'm starting to get worried about how cavalierly you treat your hands... The "spanking", the forceable opening of cartons, etc. Please be careful!
Music is related with a phenomena known as art, where we, as an emotional agent with complex general inteligence, can express ourselves through abstractions. This approach to music with AI makes the assumption that you can mathematically predict composer's feelings and musical intentions based on empirical musical data (scores and performances) and its relationships, using a powerful function aproximator called neural network. The problem is that these people are missing the most important data, which is the data that our brain uses to give meaning to music, and that composers use to compose, and for that you basically need to be able to reach human level intelligent robots.
Creative abstraction is something that is being explored but it is something innate actually and at present something only can be done in a limited fashion by AI implications
@@maxwellsim76 Thanks, nice answer. If it's innate then it's curious the complexity of Schoemberg and other state-of-the-art composers, it seems you need to do a lot of rational work to understand what they are trying to say there, or better said, to create your own version of what happens in their pieces. I keep thinking this complexity of art is only achivable by (at least) human level intelligence. Also, we do not want human level intelligent robots around, so leave art to humans please!!!