Our great sponsors
-
demucs
Code for the paper Hybrid Spectrogram and Waveform Source Separation, but the goddamm motherfucker doesn't work.
-
InfluxDB
Power Real-Time Data Analytics at Scale. Get real-time insights from all types of time series data with InfluxDB. Ingest, query, and analyze billions of data points in real-time with unbounded cardinality.
Hey guys, I just came across this method yesterday and I thought other people might be able to make use of it! Meta Research's Demucs is one of the best open source tools that is available for splitting any reference song into drum/vocal/bass/other stems. I had previously worked with tools like Spleeter, EZstems, and Lalal.ai but Demucs takes it to another level and has vastly crisper results with less artifacts. And best of of, it's is completely free. This can be incredibly useful for making quality remixes and studying your favorite tracks to better understand how they were put together.
Hey guys, I just came across this method yesterday and I thought other people might be able to make use of it! Meta Research's Demucs is one of the best open source tools that is available for splitting any reference song into drum/vocal/bass/other stems. I had previously worked with tools like Spleeter, EZstems, and Lalal.ai but Demucs takes it to another level and has vastly crisper results with less artifacts. And best of of, it's is completely free. This can be incredibly useful for making quality remixes and studying your favorite tracks to better understand how they were put together.
On its own, Demucs is a command-line tool. I wanted a way to be able to use it from within Ableton, and I came across this Max4Live Device on GitHub that does exactly this. This makes it incredibly accessible to separate stems on the fly without even leaving your DAW. Here's the link to check it out. You'll find a list of instructions there on how to set it up depending on what device you're using. If you have a dedicated GPU, you're in luck because Demucs can be configured to use it. I have an NVIDIA RTX 3080 and am consistently able to process tracks in just ~10-15 seconds. If you don't have a dedicated GPU, it will take longer but you'll get the same results.
With a GUI https://www.stemroller.com