Hacker News Reader

Comment by visarga | Hacker News Reader

> Generative music has been around for half a century,

If you start by referring to results from 50 years ago, have you tried listening to state of the art generative music systems lately? They can probably compose music better than 99% of humans.

View on HN

But we mostly listen to music written by humans who are better at writing music than 99.9999% of humans.

Yes. And there was a time when we literally used paintings to assess progress in a mine. Cars didn't outperform horse carriages for certainly 10, arguably 30 years after their invention.

This "human music" > "ai music" will flip. Suddenly. And it will never flip back.

> This "human music" > "ai music" will flip. Suddenly. And it will never flip back.

Already starting to happen with ai lyrics I use for inspiration in creating EDM music ( i.e. https://TheseLyricsDoNotExist.com/ )

This shares the same foundation as the argument that ebooks will kill physical sales and solent will change how people see food, namely that we're all purely motivated by boiling every need we have down to the most fundamental version.

It never seems to play out that way at population scales

Have you been moved by any of that music though? Am I missing something?

Yeah, it's very moving to see a human-made machine do such wonders. Fills me with awe, appreciation, and hope.

That's exactly it, though. This stuff is interesting because of the novelty of AI. The works themselves are not independently relevant (not yet, at least).

Elsewhere someone replied that art is interesting in a large part because of the personal story. How is this differemt?

99% of the time, I don't listen to music for the personal story of the artists involved. In fact, a lot of the music I listen to is made by artists that I know very little about.

Listen to some samples:

- https://openai.com/blog/jukebox/ (2020, quite good, but no classical music)

- https://openai.com/blog/musenet/ (2019 so not as good as the 2020 one, but showcases classical music)

There is no reason to assume that one cannot be moved by AI-generated music, as the AI has learnt from human-generated music and tries to mimick the styles.

While it's technically impressive and has a decent surface-level resemblance, none of the samples had any sense of direction or substance.

I can see this kind of tech taking over stuff like stock music that's automatically added to consumer holiday videos or played on the phone while you wait for a customer service agent.

That said, I'd expect the agent to be an AI long before generated music becomes independently musically relevant.

Yes - the older music is better, because it was an exploration of nondeterminism in art, and not automated replication.

Doing what has already been done is rarely compelling.

Is there anything you'd recommend for SOTA music gen?

Here are some links:

[1] https://magenta.tensorflow.org/music-transformer

[2] https://storage.googleapis.com/magentadata/papers/music-tran...

Reply on news.ycombinator.com