© 2023 Neutone Inc.

Demucs.StemSeparation

Type :
stereo-stereo
Plugin version :
v1.3.0
Demucs.StemSeparation v1.0.0 Experimental by Bogdan Teleaga Stem separation model based on Demucs that can split an audio mix into bass, drums, vocals, other using the four knobs. Since we only output one stereo channel at a time, four different instances of the plugin need to be loaded to extract every stem at the same time.

Audio Sample

Audio Sample #1
Input : Output :
Audio Sample #2
Input : Output :
Audio Sample #3
Input : Output :
Technical Information
The internal model is a retrained, smaller version of the hybrid demucs version hdemucs_mmi. The model is trained with a short segment size to exhibit better behavior with the short input buffers coming in from the DAW at inference time. Overlap add is used to merge these input buffers.
Paper Code Blog
References
Défossez, A. (2021). Hybrid spectrogram and waveform source separation. arXiv preprint arXiv:2111.03600.

Related models

Input : Output :
Type :
stereo-stereo
clipper v1.0.0 by Andrew Fyfe Audio clipper. Input : Output : Experimental
Type :
mono-mono
DDSP.sax v1.3.0 by Naotake Masuda DDSP model trained on saxophone sounds Input : Output : Experimental
Type :
mono-mono
DDSP.shakuhachi v1.3.0 by Naotake Masuda DDSP model trained on shakuhachi sounds Input : Output : Experimental
Type :
mono-mono
DDSP.violin v1.3.0 by Naotake Masuda DDSP model trained on violin sounds Input : Output : Experimental
Type :
mono-mono
ddspXsynth v0.1.0 by David Südholt DDSP cross-synthesis model trained on synth data Input : Output : Experimental
Type :
mono-mono
ddspXviolin v0.1.0 by David Südholt DDSP cross-synthesis model trained on violin data Input : Output :
Type :
mono-mono
NeuraFuzz v1.0.0 by Marco Comunità Neural fuzz + Ring modulator Input : Output :
Type :
stereo-stereo
conv1d-overdrive.random v1.0.0 by Nao Tokui Neural distortion/overdrive effect Input : Output :
Type :
mono-mono
RAVE.amen v1.3.0 by Naotake Masuda RAVE timbre transfer model trained on the Amen drum break. Input : Output : Experimental
Type :
mono-mono
RAVE.bird v1.3.0 by Nao Tokui RAVE timbre transfer model trained on field recordings of bird songs. Input : Output :
Type :
stereo-stereo
RAVE.BuddhistChant v1.2.0 by Nao Tokui RAVE timbre transfer model trained on Buddhist chant. Best with soft, sustained sounds in relatively low range. Input : Output :
Type :
stereo-stereo
RAVE.BulgarianFemaleChoir v1.2.0 by Nao Tokui Stereo RAVE timbre transfer model trained on Bulgarian female choir. Input : Output :
Type :
stereo-stereo
RAVE.choir v1.3.0 by Nao Tokui Stereo RAVE timbre transfer model trained on choir sounds. Input : Output :
Type :
mono-mono
RAVE.drumkit v1.3.0 by Naotake Masuda RAVE timbre transfer model trained on drum performances.. Input : Output :
Type :
mono-mono
RAVE.evoice v1.3.0 by Antoine Caillon RAVE model trained on English speech. Input : Output :
Type :
mono-mono
RAVE.jvoice v1.3.0 by Nao Tokui RAVE model trained on Japanese speech. Input : Output :
Type :
stereo-stereo
RAVE.kora v1.3.0 by Qosmo RAVE timbre transfer model trained on kora (african lute/harp) Input : Output :
Type :
stereo-stereo
RAVE.marimba v1.3.0 by Nao Tokui Stereo RAVE model trained on marimba sounds. Input : Output :
Type :
mono-mono
RAVE.megalodon v1.3.0 by Nao Tokui RAVE model that generates beastly bass sounds. Input : Output : Experimental
Type :
mono-mono
RAVE.MultiVox v0.5.0 by Nao Tokui AI model that converts any incoming sounds into voice sounds. Input : Output :
Type :
mono-mono
RAVE.NASA v1.3.0 by Antoine Caillon RAVE model trained on NASA recordings Input : Output :
Type :
mono-mono
RAVE.taiko v1.1.0 by Nao Tokui RAVE 2 model trained on Taiko (Japanese drums). Input : Output :
Type :
mono-mono
RAVE.VintageRadio v1.3.0 by Antoine Caillon RAVE model trained on vintage radio songs Input : Output :
Type :
stereo-stereo
RAVE.violin v1.3.0 by Naotake Masuda Stereo RAVE timbre transfer model trained on violin performances. Input : Output : Experimental
Type :
stereo-stereo
temporalconv.reverb v1.0.0 by Nao Tokui Temporal Convolution Reverb Input : Output : Experimental
Type :
stereo-stereo
Demucs.StemSeparation v1.0.0 by Bogdan Teleaga Stem separation model that can split an audio mix into bass, drums, vocals, other using the four knobs. Input : Output : Experimental
Type :
mono-mono
RAVE.djembe v1.0.0 by Nao Tokui AI model that converts any incoming audio into djembe percussion sound. Input : Output : Experimental
Type :
mono-mono
RAVE.glassharp v1.0.0 by Nao Tokui AI model that converts any incoming audio into glass harp sound.
View all models