top of page
Sky night.jpeg
  • Writer's pictureNovBee NovBee

MuseBee - AI Audio Extractor user guide

MuseBee, our AI Audio Stem Extractor, incorporates Demucs4 and KUIELAB-MDX-Net. Demucs4 is an advanced AI model designed for music source separation, and it represents the latest evolution in the Demucs series developed by Facebook AI Research (FAIR). On the other hand, KUIELAB-MDX-Net is an innovative neural network model crafted specifically for the Music Demixing Challenge (MDX) hosted on AIcrowd. Both models could accurately extract individual components from a song, such as vocals, drums, and instruments. Demucs4 accomplishes this through the implementation of its advanced deep neural networks, whereas KUIELAB-MDX-Net applies cutting-edge machine-learning techniques. With the help of the two resources, MuseBee provides a high-accuracy audio extractor.

Our tool extracts vocal tracks and instruments, including bass and drum tracks, from any given audio or song. To access our service, you can simply buy a token. The tokens are available under the Flexi Plan on our website, allowing you to buy only per the required minutes for your project. A token ID will be sent to your email for verification and use upon purchasing. Prior to initiating the separation process, you'll need to enter this token ID to confirm its validity and ensure you have sufficient time remaining for the task at hand. Following this, you're all set to upload your audio file and select the specific stem you wish to extract. The whole extraction process only takes a few minutes, and the results are promptly available. Once the extraction is complete, you can effortlessly download the separated files. Additionally, a download link and an update on the remaining seconds of your token will be sent to your email for convenience.

Recent Posts

See All


bottom of page