{"url":"/dataset/musdb18","name":"MUSDB18","full_name":null,"description_markdown":"The **MUSDB18** is a dataset of 150 full lengths music tracks (~10h duration) of different genres along with their isolated drums, bass, vocals and others stems.\r\n\r\nThe dataset is split into training and test sets with 100 and 50 songs, respectively. All signals are stereophonic and encoded at 44.1kHz.\r\n\r\nSource: [https://sigsep.github.io/datasets/musdb.html#musdb18-compressed-stems](https://sigsep.github.io/datasets/musdb.html#musdb18-compressed-stems)\r\nImage Source: [https://sigsep.github.io/datasets/musdb.html#musdb18-compressed-stems](https://sigsep.github.io/datasets/musdb.html#musdb18-compressed-stems)","description_withheld":null,"homepage":"https://sigsep.github.io/datasets/musdb.html","introduced_date":null,"introduced_date_note":null,"introduced_by":{"paper":null,"title":"The MUSDB18 corpus for music separation","first_author":null,"url":"https://doi.org/10.5281/zenodo.1117372"},"license":{"name":"Various (see link)","url":"https://sigsep.github.io/datasets/musdb.html"},"modalities":[{"name":"Audio","url":"/datasets/modality/audio"}],"tasks":[{"name":"Audio Source Separation","url":"/task/audio-source-separation","datasets_with_task":"/datasets/task/audio-source-separation"},{"name":"Music Source Separation","url":"/task/music-source-separation","datasets_with_task":"/datasets/task/music-source-separation"},{"name":"Cadenza 1 - Task 1 - Headphone","url":"/task/cadenza-1-task-1-headphone","datasets_with_task":"/datasets/task/cadenza-1-task-1-headphone"}],"languages":[],"variants":["MUSDB18"],"data_loaders":[{"repo":"https://github.com/pytorch/audio","url":"https://pytorch.org/audio/stable/generated/torchaudio.datasets.MUSDB_HQ.html#torchaudio.datasets.MUSDB_HQ","frameworks":["pytorch"]},{"repo":"https://github.com/sigsep/open-unmix-pytorch","url":"https://github.com/sigsep/open-unmix-pytorch","frameworks":["pytorch"]},{"repo":"https://github.com/sigsep/sigsep-mus-db","url":"https://sigsep.github.io/sigsep-mus-db/","frameworks":[]}],"num_papers_in_archive":106,"source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28"},"benchmarks":[{"leaderboard":"/sota/music-source-separation-on-musdb18","task":"Music Source Separation","dataset_variant":"MUSDB18","rows":27,"metrics":["SDR (avg)","SDR (vocals)","SDR (drums)","SDR (bass)","SDR (other)"],"first_row_in_archive_order":{"model":"Sparse HT Demucs (fine tuned)","paper":"/paper/hybrid-transformers-for-music-source","metrics":{"SDR (avg)":"9.20","SDR (bass)":"10.47","SDR (drums)":"10.83","SDR (other)":"6.41","SDR (vocals)":"9.37"},"code_links":[{"title":"facebookresearch/demucs","url":"https://github.com/facebookresearch/demucs"},{"title":"zhaozhipeng1997/demucs_ascend910_pytorch","url":"https://github.com/zhaozhipeng1997/demucs_ascend910_pytorch"}]},"note":"rows are the archive's own order at snapshot; nothing here re-ranks them"},{"leaderboard":"/sota/cadenza-1-task-1-headphone-on-musdb18","task":"Cadenza 1 - Task 1 - Headphone","dataset_variant":"MUSDB18","rows":2,"metrics":["HAAQI"],"first_row_in_archive_order":{"model":"Baseline Demucs","paper":"/paper/the-first-cadenza-signal-processing-challenge","metrics":{"HAAQI":"0.2592"},"code_links":[{"title":"claritychallenge/clarity","url":"https://github.com/claritychallenge/clarity/tree/main/recipes/cad1/task1"}]},"note":"rows are the archive's own order at snapshot; nothing here re-ranks them"}],"papers_with_a_benchmark_row":[{"paper":"/paper/the-first-cadenza-signal-processing-challenge","title":"The First Cadenza Signal Processing Challenge: Improving Music for Those With a Hearing Loss","date":"2023-10-09","rows_on_this_dataset":2,"code_links":1,"syntology":null},{"paper":"/paper/contrastive-learning-based-deep-latent","title":"Contrastive Learning based Deep Latent Masking for Music Source Separation","date":"2023-08-20","rows_on_this_dataset":1,"code_links":0,"syntology":null},{"paper":"/paper/sound-demixing-challenge-2023-music-demixing","title":"Sound Demixing Challenge 2023 Music Demixing Track Technical Report: TFC-TDF-UNet v3","date":"2023-06-15","rows_on_this_dataset":1,"code_links":1,"syntology":null},{"paper":"/paper/hybrid-transformers-for-music-source","title":"Hybrid Transformers for Music Source Separation","date":"2022-11-15","rows_on_this_dataset":2,"code_links":2,"syntology":{"read_at":"2026-09-24T18:15:14+00:00","samples_harvested":10,"samples_ran":0,"samples_unverified":10,"pointer_only_for_licence":10,"claim":"Per-sample execution on synthesized fixtures; not a correctness claim."}},{"paper":"/paper/an-efficient-short-time-discrete-cosine","title":"An Efficient Short-Time Discrete Cosine Transform and Attentive MultiResUNet Framework for Music Source Separation","date":"2022-11-14","rows_on_this_dataset":1,"code_links":1,"syntology":null},{"paper":"/paper/music-source-separation-with-band-split-rnn","title":"Music Source Separation with Band-split RNN","date":"2022-09-30","rows_on_this_dataset":2,"code_links":3,"syntology":{"read_at":"2026-09-24T18:15:14+00:00","samples_harvested":10,"samples_ran":0,"samples_unverified":10,"pointer_only_for_licence":0,"claim":"Per-sample execution on synthesized fixtures; not a correctness claim."}},{"paper":"/paper/hierarchic-temporal-convolutional-network","title":"Hierarchic Temporal Convolutional Network With Cross-Domain Encoder for Music Source Separation","date":"2022-06-30","rows_on_this_dataset":1,"code_links":0,"syntology":null},{"paper":"/paper/cws-presunet-music-source-separation-with","title":"CWS-PResUNet: Music Source Separation with Channel-wise Subband Phase-aware ResUNet","date":"2021-12-09","rows_on_this_dataset":1,"code_links":1,"syntology":null},{"paper":"/paper/kuielab-mdx-net-a-two-stream-neural-network","title":"KUIELab-MDX-Net: A Two-Stream Neural Network for Music Demixing","date":"2021-11-24","rows_on_this_dataset":1,"code_links":1,"syntology":null},{"paper":"/paper/hybrid-spectrogram-and-waveform-source","title":"Hybrid Spectrogram and Waveform Source Separation","date":"2021-11-05","rows_on_this_dataset":1,"code_links":1,"syntology":null},{"paper":"/paper/lasaft-latent-source-attentive-frequency","title":"LaSAFT: Latent Source Attentive Frequency Transformation for Conditioned Source Separation","date":"2020-10-22","rows_on_this_dataset":1,"code_links":1,"syntology":{"read_at":"2026-09-24T18:15:14+00:00","samples_harvested":3,"samples_ran":1,"samples_unverified":2,"pointer_only_for_licence":0,"claim":"Per-sample execution on synthesized fixtures; not a correctness claim."}},{"paper":"/paper/all-for-one-and-one-for-all-improving-music","title":"All for One and One for All: Improving Music Separation by Bridging Networks","date":"2020-10-08","rows_on_this_dataset":1,"code_links":5,"syntology":{"read_at":"2026-09-24T18:15:14+00:00","samples_harvested":10,"samples_ran":1,"samples_unverified":9,"pointer_only_for_licence":0,"claim":"Per-sample execution on synthesized fixtures; not a correctness claim."}},{"paper":"/paper/d3net-densely-connected-multidilated-densenet","title":"D3Net: Densely connected multidilated DenseNet for music source separation","date":"2020-10-05","rows_on_this_dataset":2,"code_links":1,"syntology":null},{"paper":"/paper/meta-learning-extractors-for-music-source","title":"Meta-learning Extractors for Music Source Separation","date":"2020-02-17","rows_on_this_dataset":1,"code_links":1,"syntology":null},{"paper":"/paper/music-source-separation-in-the-waveform-1","title":"Music Source Separation in the Waveform Domain","date":"2019-11-27","rows_on_this_dataset":2,"code_links":1,"syntology":null},{"paper":"/paper/spleeter-a-fast-and-state-of-the-art-music","title":"Spleeter: A Fast And State-of-the Art Music Source Separation Tool With Pre-trained Models","date":"2019-11-04","rows_on_this_dataset":1,"code_links":3,"syntology":null},{"paper":"/paper/tf-attention-net-an-end-to-end-neural-network","title":"Sams-Net: A Sliced Attention-based Neural Network for Music Source Separation","date":"2019-09-12","rows_on_this_dataset":1,"code_links":1,"syntology":null},{"paper":"/paper/open-unmix-a-reference-implementation-for","title":"Open-Unmix - A Reference Implementation for Music Source Separation","date":"2019-09-08","rows_on_this_dataset":2,"code_links":1,"syntology":null},{"paper":"/paper/end-to-end-music-source-separation-is-it","title":"End-to-end music source separation: is it possible in the waveform domain?","date":"2018-10-29","rows_on_this_dataset":1,"code_links":2,"syntology":{"read_at":"2026-09-24T18:15:14+00:00","samples_harvested":1,"samples_ran":1,"samples_unverified":0,"pointer_only_for_licence":1,"claim":"Per-sample execution on synthesized fixtures; not a correctness claim."}},{"paper":"/paper/tasnet-surpassing-ideal-time-frequency","title":"Conv-TasNet: Surpassing Ideal Time-Frequency Magnitude Masking for Speech Separation","date":"2018-09-20","rows_on_this_dataset":2,"code_links":17,"syntology":{"read_at":"2026-09-24T18:15:14+00:00","samples_harvested":36,"samples_ran":7,"samples_unverified":29,"pointer_only_for_licence":16,"claim":"Per-sample execution on synthesized fixtures; not a correctness claim."}},{"paper":"/paper/wave-u-net-a-multi-scale-neural-network-for","title":"Wave-U-Net: A Multi-Scale Neural Network for End-to-End Audio Source Separation","date":"2018-06-08","rows_on_this_dataset":1,"code_links":10,"syntology":{"read_at":"2026-09-24T18:15:14+00:00","samples_harvested":19,"samples_ran":1,"samples_unverified":18,"pointer_only_for_licence":1,"claim":"Per-sample execution on synthesized fixtures; not a correctness claim."}},{"paper":"/paper/mmdenselstm-an-efficient-combination-of","title":"MMDenseLSTM: An efficient combination of convolutional and recurrent neural networks for audio source separation","date":null,"rows_on_this_dataset":1,"code_links":1,"syntology":{"read_at":"2026-09-24T18:15:14+00:00","samples_harvested":7,"samples_ran":0,"samples_unverified":7,"pointer_only_for_licence":0,"claim":"Per-sample execution on synthesized fixtures; not a correctness claim."}}],"syntology_totals":{"read_at":"2026-09-24T18:15:14+00:00","papers_with_samples":8,"samples_harvested":96,"samples_ran":11,"samples_unverified":85,"pointer_only_for_licence":28,"papers_with_no_sample_that_ran":3,"note":"the per-paper counts above, summed; not a rate"},"papers_note":"The archive never published its papers-using-dataset list; these are papers with a leaderboard row on this dataset's benchmarks."}