BS Roformer SW model, which generates 6 stems at once with superior quality.

Algorithm Demucs4 HT. It's fast and gives relatively good quality for bass/drums/other stems.

BS Roformer model. Excellent quality for vocals/instrumental separation.

BS PolarFormer is a high-quality algorithm for vocal separation.

Algorithm for separating tracks into vocal and instrumental parts based on the MelBand Roformer neural network

Set of MDX23C models which is based on code released by kuielab for Sound Demixing Challenge 2023. Very good for vocals/instrumental separation.

Algorithm for separating tracks into vocal and instrumental parts based on the SCNet neural network

MDX B models are based on kuielab code from Music Demixing Challenge 2021. Models were retrained by UVR team on big dataset. For long time models were best for vocals/instrumental separation.

A set of models from the Ultimate Vocal Remover program, which are based on the old VR architecture. Most of the models are vocal, but there are also special models for karaoke, piano, removing reverberation effects, etc.

Demucs4 Vocals 2023 model - it's Demucs4 HT model fine-tuned on big vocals dataset.

Algorithm for extracting only lead vocals and everything else based on the MelBand Roformer and SCNet models.

The MDX-B Karaoke model was prepared as part of the Ultimate Vocal Remover project. The model produces high-quality lead vocal extraction from a music track.

An unique model for removing crowd sounds from music recordings (applause, clapping, whistling, noise, laugh etc.).

Medley Vox is an algorithm for separating multiple singers within a single music track and evaluation dataset for this task.

MVSep Multichannel BS - uses the best vocal model to extract sound from multi-channel audio (5.1, 7.1, etc.).

A model for separating male and female voices within a single vocal track. The track should contain only voices, no music.

Choir Extraction Model

Model to separate vocals and strings to SATB parts (Soprano, Alto, Tenor, and Bass)

The MVSep Drums model produces high-quality separation of music into a drums part and everything else.

The MVSep Bass model produces high-quality separation of music into a bass part and everything else.

Synth extraction model

The DrumSep model divides the drum track into several types: 'kick', 'snare', 'toms', 'cymbals' (it includes 'hh', 'ride', 'crash').

MVSep Piano model is based on MDX23C, MelRoformer and SCNet Large architectures. It produces high quality separation for piano and other stems.

No data found

The MVSep Keys is a high quality model for separating music into keys instruments and everything else.

The MVSep Organ model produces high-quality separation of music into an organ part and everything else.

No data found

No data found

No data found

No data found

The Rhodes (Fender Rhodes) is an iconic electromechanical keyboard instrument designed by Harold Rhodes.

The Clavinet is an electromechanical keyboard instrument developed and manufactured by the German company Hohner from 1964 to 1982.

Mellotron is a electromechanical keyboard instrument developed in England in the early 1960s.

The MVSep Guitar model produces high-quality separation of music into a guitar part (including acoustic and electronic) and everything else.

No data found

No data found

No data found

Algorithm for isolating pedal steel guitar sounds from a melody

The MVSep Plucked Strings is a high quality model for separating music into plucked strings instruments and everything else.

No data found

No data found

No data found

No data found

No data found

No data found

The MVSep Bowed Strings is a high quality model for separating music into bowed string instruments and everything else.

No data found

No data found

No data found

No data found

The MVSep Wind model produces high-quality separation of music into a wind part and everything else.

The MVSep Brass is a high quality model for separating music into brass wind instruments and everything else.

The MVSep Woodwind is a high quality model for separating music into woodwind instruments and everything else.

No data found

No data found

No data found

No data found

No data found

No data found

No data found

No data found

No data found

No data found

The MVSep Bagpipes model provides high-quality music separation, isolating the bagpipes from the rest of the track.

No data found

The MVSep Percussion is a high quality model for separating music into percussion instruments and everything else.

No data found

No data found

No data found

No data found

No data found

No data found

No data found

No data found

No data found

No data found

No data found

No data found

No data found

BandIt Plus model for separating tracks into speech, music and effects.

Bandit v2 is a model for cinematic audio source separation in 3 stems: speech, music, effects/sfx. It was trained on DnR v3 dataset.

MVSep DnR v3 is a cinematic model for splitting tracks into 3 stems: music, sfx and speech.

The MVSep Braam model provides high-quality music separation, isolating the cinematic "Braam" sound effect from the rest of the track.

Algorithm for isolating "Riser" sounds from a musical melody

No data found

The algorithm restores the quality of audio. For example MP3 files compressed to 128 kbps or lower and other types.

Set of different models to remove reverberation effect from music.

No data found

Algorithm AudioSR: Versatile Audio Super-resolution at Scale. Algorithm restores high frequencies.

FlashSR - audio super resolution algorithm for restoring high frequencies

Generating audio based on a given text prompt

Whisper is a pre-trained model for automatic speech recognition (ASR) and speech translation.

Parakeet by NVIDIA is a state-of-the-art automatic speech recognition (ASR) model designed for accurate and efficient conversion of spoken language into text.

VibeVoice - is a model for generating natural conversational dialogues from text with the ability to use a reference voice for cloning purposes.

VibeVoice (TTS) - is a model for generating natural conversational dialogues from text, capable of creating dialogues with up to 4 speakers and durations of up to 90 minutes.

Qwen3-TTS (Custom Voice) is a powerful speech generation model with predefined speakers

Qwen3-TTS (Voice Design) is a powerful speech generation model capable of creating a voice based on a description.

Qwen3-TTS (Voice Cloning) is a powerful speech generation model capable of cloning a voice based on a reference audio file

A model for separating a musical composition into 53 different musical instruments

Advanced AI Model for Stereo-to-Surround Upmixing

A model that recovers pure dry lead vocals from processed audio

Bark - is a generative model for realistic speech, music, and emotions.

MVSep MultiSpeaker (MDX23C) - this model tries to isolate the most loud voice from all other voices.

The algorithm adds "whispering" effect to vocals.

Algorithm for extracting the phantom center from a audio track

Matchering is a novel tool for audio matching and mastering.

SOME is a MIDI extractor that can convert singing voice to MIDI sequence.

High-quality transcription of piano music into MIDI

Basic Pitch is a modern neural network that converts melodic audio recordings into notes (MIDI format).

Automatic transcription of drum audio to MIDI

HeartMuLa - an open-source AI model for music generation (an alternative to Suno and Udio)

Algorithm Demucs3 (A and B versions)

No data found

Experimental model VitLarge23 based on Vision Transformers. In terms of metrics, it is slightly inferior to the MDX23C, but may work better in some cases.

No data found

No data found

No data found

No data found

No data found

No data found

No data found

No data found

No data found

The LarsNet model divides the drums stem into 5 types: 'kick', 'snare', 'cymbals', 'toms', 'hihat'.