Demucs is a deep learning model for music source separation, which is the task of separating individual audio sources from a music mixture. The model is trained on a large dataset of mixed music signals and their corresponding isolated sources. It learns to separate different sources by predicting the masks that indicate which parts of the mixture belong to each source. Demucs has achieved state-of-the-art performance in various source separation benchmarks and can be used for applications such as remixing, analysis, and enhancement of music recordings.

Use cases

Demucs, a deep learning model for music source separation, opens up a multitude of possibilities for the technical community. One potential use case for Demucs is in the field of remixing, where it can unravel the individual audio sources from a music mixture, allowing DJs and producers to isolate and manipulate specific elements of a track with greater precision and control. This AI model could also find value in music analysis, as it can separate different sources and provide insights into the composition and arrangement of a piece. Furthermore, Demucs can be employed for the enhancement of music recordings, enabling the removal of unwanted noise or background elements to improve the overall audio quality. With its state-of-the-art performance, this model has shown significant potential and could potentially be integrated into products such as music production software, audio editing tools, or even music streaming platforms to offer users new and exciting ways to interact with and experience music.



Cost per run
Avg run time
Nvidia T4 GPU

Creator Models

Pix2pix Zero$?4,206
Night Enhancement$0.0104520,721
Mindall E$?1,645
Compositional Vsual Generation With Composable Diffusion Models Pytorch$0.01155774

Similar Models

Try it!

You can use this area to play around with demo applications that incorporate the Demucs model. These demos are maintained and hosted externally by third-party creators. If you see an error, message me on Twitter.

Currently, there are no demos available for this model.


Summary of this model and related resources.

Model NameDemucs
Demucs Music Source Separation
Model LinkView on Replicate
API SpecView on Replicate
Github LinkView on Github
Paper LinkView on Arxiv


How popular is this model, by number of runs? How popular is the creator, by the sum of all their runs?

Model Rank
Creator Rank


How much does it cost to run this model? How long, on average, does it take to complete a run?

Cost per Run$-
Prediction HardwareNvidia T4 GPU
Average Completion Time-