One comment

  1. admin

    on music and machine learning:
    machine learning and notes:
    check out:
    check out:
    check out:
    check out:
    check out:

    Cute fish w/ great visual feedback, like the idea, well executed, good that it presents itself as being complete.


    Should provide a clearer comparison between the ‘fish out of the pond’ and the ‘well trained fish’

    Smashing head against star reflections — poetic idea
    also inadvertently interesting commentary about machine learning/art vs. human learning/art. computers can only reflect back what humans originally create?

    Very cute.
    Needs a better (more legible) user interface for specifying whether the user thinks it’s good/bad.

    There should be one note that sounds absolutely horrible so that you can train it to avoid that note at all cost← just for documentation purposes

    Documentation should highlight contrast between not well trained and well trained fish.

    I’m confused about what about the music is represented by the machine learning algorithm.

    In wonder if this could be a bi-directional test, where the user is being trained like a game of Simon.

    You need to be more clear about what you’re representing: pitch sequences
    And what you’re not representing: timbre, rhythm, melodic shape, consonance/dissonance, resolution to a tonic.

    Musicality = correctness? Or are you training for correctness

    Adorable, maybe making instructions easily accessible for people to read would help? Would be confusing to use otherwise

    very soothing soundtrack this seems like an iphone app or a galaxy background etc.

    Love the intimate scale of this;it’s just between you and that fish

    I think it may help people who are “playing with the application” to be able to see the notes, so they can compare the pattern as well.

    I agree that what the fish is learning needs to be clarified more.