正在加载视频...

视频加载失败

The point is: compositions are written when they're published(lyrics+music)...there are literally sheet music websites that generate royalties for songwriters Also Beyoncé can sight read & chart music on paper- which was a prerequisite to be admitted in the HSPVA's vocal program

27,714 次观看 • 4 个月前 •via X (Twitter)

0 条评论

暂无评论

原始帖子的评论将显示在这里

相关视频

Suno is limiting downloads, basically making a lot of the songs you make locked in their 'walled garden'. One of their main competitors, Udio, did the same thing after a settlement with UMG that restricts any downloads. This is 100% an injunction from the music labels who want to limit the "damage" that AI does to the profitability of music. There are two viable alternatives... >Minimax music 3.0. It is currently free to use up to 500 songs a day as long as your account qualifies for beta testing their music model, which should be anyone. It is a closed-weight model, but there are no restrictions on what you download and how much you download. The music quality is not as good as Udio or Suno, but it's pretty close overall, and it can understand very complex prompts to guide music generation. >Open-source music is not a very good alternative... but right now SongGeneration-LeVo2 is the best AI music model that's open-source right now. The music I put on the post is one such generation that I created on my PC. The problem is that it's hard to use. The comfyUI version requires installation on a separate portable install, specific transformer and flash attention versions, and additional libraries installed. The required models and checkpoints are scattered around Hugging Face, and some of the links from the original repo are dead. There are not many English guides, and most of the training data is in Chinese, so it struggles with English lyrics. To some extent, the demand for music generation isn't that high, compared to images and videos, so overall not many alternatives are around.

Emerald Apple

20,420 次观看 • 1 个月前