Israeli AI can write you a cadenza

Israeli AI can write you a cadenza


norman lebrecht

November 14, 2020

From the Jerusalem Post:

A team of Israeli researchers won the “Best Research” award at the 21st International Society for Music Information Retrieval conference (ISMIR 2020) on Tuesday for their paper titled, “BebopNet: Deep Neural Models for Personalized Jazz Improvisation.
Authored by M.Sc. students Nadav Bhonker and Shunit Haviv Hakimi, along with their adviser Prof. Ran El-Yaniv at the Henry and Marilyn Taub Faculty of Computer Science at the Technion-Institute of Technology, the paper indicates that it is possible to model and optimize personalized jazz preferences.
If they can do jazz improvisation, how long before we get a new cadenza for the Beethoven violin concerto?


  • christopher storey says:

    It’s Bhonkers !

  • Pianofortissimo says:

    The cadenza played in the recording whose cover art illustrates this blog post is horrible, mixing Beethoven’s cadenza for the pianoversion with electronics. Tha’s typical of modernist Kremer but I cannot conceive of Harnoncourt accepting that cadenza.

    The same cadenza played only on the violin (Schneiderhan, Ricci) is wonderful.

  • Pianofortissimo says:

    AI-produced art: the advent of the real ‘Entartete Kunst.’

  • J Kaznowski says:

    A misleading title. Should at least have a question mark after it.

  • Nijinsky says:


    Why not go the whole way and have computerized celebrities made up of AI and computerized whatever, and drones or robots….

    I find this just amazing. With computers one can actually delve in areas that one couldn’t before. Mozart then wouldn’t have to hold a whole piece in his head before writing it down, given that once it’s down you can’t change it; instead you can alter whatever you want and the program will adjust, and you can print out the new version without having to rewrite everything. A whole new arena of thoughtfulness. Same with the written word and movie editing and I don’t know what else I can’t come up with being so irritated at what would seem to be a just would one take it seriously.

    An AI Beethoven concerto cadenza!?

    You can also with modern digital equipment change the tuning. There’s 100 cents between each half tone, and you can in real time send message to change the tuning of the synthesizer, in essence liberating the pitch or pitches going on from the 12 notes that have been around limiting pitch since the well tempered system was devised so fingers could traverse the keyboard without getting lost beyond 12 keys to play on each octave. Two pure minor sevenths on top of each other and you’re inbetween two well tempered pitches.

    But no, let’s see if AI can come up with a cadenza, the whole classical world having lost the ability to write their own cadenzas mostly, or that such is even taught or encouraged at any conservatory.

    Well, that’s goes well with the whole schpiel. WHY NOT!? The paint by number nice music, or the paint by number method, when it’s (the one “painting”) rebellious supposedly, way of producing such music, along with paint by number wars as to which method is the one that should be taught in conservatories. And then paint by number morality, and supported with paint by number evidence; all very neatly subjugated into paint by number territories based on paint by number safety.

    Get your credit card out and your cell phone and start looking for that cadenza….

    You can order tickets also, and meals to go with it, all the same fashion.

    Clothes too.

    Glasses, healthcare, politicians, all soon to be brought to you by AI as well.

    And let’s not forget religion.

    I just want to know what’s the difference between a professional AI drone and an amateur?

  • Anonymous says:

    From the paper:
    “While our generated solos are locally coherent and often interesting/pleasing, they lack the qualities of professional jazz related to general structure such as motif development and variations…The challenge of evaluating neural networks that gen- erate art remains a central issue in this research field. An ideal jazz solo should be creative, interesting and meaningful. Nevertheless, when evaluating jazz solos, there are no mathematical definitions for these properties—as yet”