Spleeter: The AI Tool That Splits a Song into Stems with Ease
Separating the parts of a song, such as the vocals or individual instruments, has always been a headache for producers, DJs and people making karaoke tracks. There are many ways to do it, but the process is hugely time-consuming and the results are often not quite what you hoped for. A new open-source AI tool called Spleeter makes the job easier than ever!
Spleeter was developed by the music streaming service Deezer for research purposes, and the company recently released it as open source, uploading it to GitHub for anyone to download and use.
All you have to do is feed an audio file into Spleeter, and it splits the song into two, four or five separate parts, which we usually call stems. The results couldn't be better, and it runs even faster if your computer has a dedicated GPU.
Tech expert Andy Baio wrote a post about Spleeter with plenty of his own examples. He noted that although the vocals sometimes sound a little robotic, the clarity is astonishing compared with other solutions.
Deezer says it tested the software on 20,000 tracks with separated vocals across many different music genres. All in all, Spleeter is a great example of how AI tools can make creative work simpler.
Deezer has no plans to turn Spleeter into a commercial, money-making product, but developers can use the source code to build apps with a more visual, easier-to-use interface. The company will also use Spleeter itself for a range of research projects to improve its streaming service.
Source: Loop Central
