zone411

38 ms·

481.
by zone411 4y ago
They might be helpful for humans, just as rhythm or a bassline might be, but it's not what makes this song great (to answer the title of the post). This would have been a great song if it was just sung without any accompaniment, even t
482.
by zone411 4y ago
The most important part is the melody. I don't buy that "chords generate the melody" as he says early in the video, it's other way around. That was my philosophy when writing my AI melody generating assistant and it&#x27
483.
by zone411 4y ago
If you connect various steps of what can already be done well, such as melody generation (my project), lyrics generation, singing voice synthesis, chord generation, and various semi-automated VST plugins with generation features, you can cr
484.
by zone411 4y ago
In my melody generation system I'm already including melodies that I've judged as "good" ( https://www.youtube.com/playlist?list=PLoCzMRqh5SkFwkumE578Y... ) in the updated training set. Since the number of
485.
by zone411 4y ago
a) Diffusion is not just used to upsample images but also to create them. b) It has seen images with descriptions of "corn," "cobra," "farm," and it has seen images of "A made out of B" and "C on
486.
by zone411 4y ago
When diffusion models are used, the inference time could be meaningful. But then this is only 64x64 with upsampling, so probably not too bad.
487.
by zone411 4y ago
50 buyins is pretty reasonable for an average (which means losing because of the rake) player for a year. You can plug in the numbers into one of the poker variance calculators. At 1 hour per day (60 hands), 10 BB/100 rake, NLHE 6-max
488.
by zone411 4y ago
Fun idea. We're about test out melodies I've created together with a generative neural net and see how they're rated computer to real melodies. The plan is use Amazon's Mechanical Turk but one problem is what to do about
489.
by zone411 4y ago
The most of obvious way would be to use the same test as previously and just compare the scores.
490.
by zone411 4y ago
Not really any new info there for anybody who has followed basic economic news but still I would take these sorts of predictions much more seriously if he offered some kind of proof that he's putting his money where his mouth is. If he
491.
by zone411 4y ago
Covid won't help: https://www.thelancet.com/journals/eclinm/article/PIIS2589-5... .
492.
America Reached One Million Covid Deaths(nytimes.com)
5 points by zone411 4y ago|0 comments
493.
by zone411 4y ago
I believe the real AI music revolution is still ahead of us. Most existing tools don't really use modern deep neural nets and the ones that have just been released and do are quite good (e.g. Synthesizer V). The project I released yest
494.
by zone411 4y ago
I found that it's best to avoid most of these webpage explanations that pop up in Google (often on Towards Data Science and Medium). You can get a better understanding by reading intro sections of actual research papers.
495.
by zone411 4y ago
Yes, I am. It does many things well but last time I used it (about a year ago), I couldn't get it to do what I focused on with this project: create a catchy melodic hook that could be a base of a hit song or an earworm commercial jingl
496.
by zone411 4y ago
Thanks! I wasn't familiar with a couple of these, I will definitely check them out.
497.
by zone411 4y ago
Thanks. I used Piano VFX (it's a bit buggy but workable) and VSDC (it's missing any sort of more advanced functions like batch rendering but does visualizations very well).
498.
by zone411 4y ago
I'm guessing the link comes from my Show HN post earlier today about a generative diffusion process neural net assistant which let me create 60 pretty catchy melodies that linked to this page ( https://news.ycombinator.com&#x
499.
by zone411 4y ago
> In a sense, the lyrics could be spoken and then applied to the .midi melody. The result would sound similar to the SoundCloud vocal OP link, but with "words". Yes, Synthesizer V Studio can do it quite well! I've added so
500.
by zone411 4y ago
Not yet. The current version is not fully automated and it requires human input, so it will take a decent amount of work and documentation to make it ready for users. It also mixes C++ with Python in a few places for performance reasons whe
501.
Show HN: Catchy melodies made with a diffusion-based neural net assistant(youtube.com)
38 points by zone411 4y ago|14 comments
502.
by zone411 4y ago
Space Time is excellent compared to Hossenfelder's videos.
503.
by zone411 4y ago
She makes videos about many topics on which she's not even close to an expert.
504.
by zone411 4y ago
Her Simulation Theory video was absolutely awful ( https://lech.substack.com/p/sabine-hossenfelders-video-the-s... ), so it's not just a gut reaction.
505.
by zone411 4y ago
Agree. I stopped watching her stuff after her really terrible video about the Simulation Hypothesis, which showed that she didn't even grasp the argument. Since she feels free to opine about things she clearly knows nothing about, I ca
506.
by zone411 5y ago
Her video on the Simulation Hypothesis was very poor https://lech.substack.com/p/sabine-hossenfelders-video-the-s... .
507.
by zone411 5y ago
Related to earlier stories about Próspera: https://news.ycombinator.com/item?id=26609598 https://news.ycombinator.com/item?id=28758212 https://news.ycombinator.com/item?id=30773772 https:&
508.
Honduras Repeals ZEDEs(laprensa.hn)
47 points by zone411 5y ago|10 comments
509.
by zone411 5y ago
That's pretty much how AI image compression works - you're describing standard autoencoder (encoder/decoder) architecture with some adaptiveness.
510.
by zone411 5y ago
Some more examples: https://twitter.com/sama/status/1511724264629678084

More ›