Spleeter now released as Open-Source – Package

Spleeter now released as Open-Source – Package

New York, November 7, 2019.

Splitting a song into individual vocals and instruments has always been a problem for producers, DJs and anyone else who wants to play with isolated audio. There are many ways to do this, but the process can be time consuming and the results often imperfect. A new open source AI tool makes this tricky task faster and easier.

open-source package

The software is called Spleeter and was developed by the music streaming service Deezer for research purposes. Yesterday, the company released it as an open source package that makes the code on Github available for everyone to download and use. Feed Spleeter with an audio file and it (spleets) splits it into two, four or five separate audio tracks known as styles. The results are not perfect, but they are excellent to use and Spleeter itself is very fast. When running on a dedicated graphics processor, it can split audio files into four sections 100 times faster than in real time.

This tool is very powerful, but be warned: you will need some technical knowledge to use it. If you don’t play regularly with software like Python or Google’s AI toolkit TensorFlow (which was trained with Spleeter), you’ll need to download some programs to get everything running. And you need to know how to use command line input (albeit a very simple one) instead of a more accessible visual interface.

Deezer notes that this is not the first time people have used machine learning to automate this task, and that the company’s successes are based on much previous research. Speaking to The Verge about email, Aurelien Herault, Chief Data and Research Officer of Deezer, says the company has trained its software on 20,000 pieces of pre-isolated vocal music in a variety of genres. From this information, the software learned how to isolate the tracks themselves.

Overall, Spleeter is another fantastic example of how AI tools can simplify tricky parts of creative work. Machine learning is currently used to automate a range of time-consuming tasks, from removing backgrounds from images to upscaling textures in old video games. And these tools are increasingly being integrated into consumer software, from Adobe Photoshop to new competitors like Runway ML.

removing backgrounds on pictures to upscaling textures

Deezer says it has no plans to turn Spleeter into a consumer tool, but others could take their work and beat a simple interface to it.

Deezer himself uses Spleeter for a number of research applications that help improve the streaming service. „Internally, we use it as a pre-processing tool for complex research tasks such as music categorization, transcription and speech recognition,“ says Herault.

Or, of course, you can just use it to get a better grip on Scatman. Ski-bi dibby dibby dibby dib yo da dub dub.