Just released Jukebox, a neural network that generates music as raw audio. All of the sound you hear, including the singing, is generated:
Introducing Jukebox, a neural net that generates music, including rudimentary singing, as raw audio in a variety of genres and artist styles. We're releasing a tool for everyone to explore the generated samples, as well as the model and code: openai.com/blog/jukebox/

Apr 30, 2020 · 4:08 PM UTC

10
47
7
282
Replying to @gdb
The colab from github has restricted permissions :)
1
2
Should be fixed!
1
4
Replying to @gdb
Could you perhaps make a way to vote up autogenerated music? perhaps with enough people voting and autogenerated songs, some really amazing creative songs could come up through this
1
Replying to @gdb
Amazing!
Replying to @gdb @Montreal_AI
This is amazing !! How about the lyrics ?!?
Replying to @gdb
Only at @OpenAI, there are songs written about Stochastic Gradient Descent 😂
4
Replying to @gdb
I think that instead of my toe sis it's trying to tell us something about mitosis. Maybe it's a cryptic description of a cure for the coronavirus as gift to humanity?
Replying to @gdb
Moving humanity forward
Replying to @gdb
Do artists have any rights to the music generated from models trained with their music?