Music Source Separation On Musdb18
평가 지표
SDR (avg)
SDR (bass)
SDR (drums)
SDR (other)
SDR (vocals)
평가 결과
이 벤치마크에서 각 모델의 성능 결과
비교 표
| 모델 이름 | SDR (avg) | SDR (bass) | SDR (drums) | SDR (other) | SDR (vocals) |
|---|---|---|---|---|---|
| end-to-end-music-source-separation-is-it | 3.5 | 2.49 | 4.60 | 0.54 | 3.46 |
| mmdenselstm-an-efficient-combination-of | 6.04 | 5.40 | 6.81 | 4.80 | 7.16 |
| kuielab-mdx-net-a-two-stream-neural-network | 7.54 | 7.86 | 7.33 | 5.95 | 9.00 |
| an-efficient-short-time-discrete-cosine | 6.81 | 5.88 | 7.63 | 5.14 | 8.57 |
| music-source-separation-in-the-waveform-1 | 6.79 | 7.60 | 7.58 | 4.69 | 7.29 |
| wave-u-net-a-multi-scale-neural-network-for | 3.23 | 3.21 | 4.22 | 2.25 | 3.25 |
| music-source-separation-in-the-waveform-1 | 6.28 | 7.01 | 6.86 | 4.42 | 6.84 |
| hybrid-spectrogram-and-waveform-source | 7.72 | 8.67 | 8.58 | 5.59 | 8.04 |
| tasnet-surpassing-ideal-time-frequency | 5.73 | 5.66 | 6.08 | 4.37 | 6.81 |
| music-source-separation-with-band-split-rnn | 8.97 | 8.16 | 10.15 | 7.08 | 10.47 |
| spleeter-a-fast-and-state-of-the-art-music | 5.91 | 5.51 | 6.71 | 4.02 | 6.86 |
| sound-demixing-challenge-2023-music-demixing | 8.34 | 8.45 | 8.44 | 6.86 | 9.59 |
| tasnet-surpassing-ideal-time-frequency | 6.32 | 7.00 | 7.11 | - | 6.74 |
| open-unmix-a-reference-implementation-for | 6.316 | 6.015 | 7.148 | 4.889 | 7.213 |
| hierarchic-temporal-convolutional-network | 6.89 | 7.92 | 7.33 | 4.92 | 7.37 |
| meta-learning-extractors-for-music-source | 5.52 | 5.58 | 5.91 | 4.19 | 6.40 |
| cws-presunet-music-source-separation-with | 6.77 | 5.93 | 6.38 | 5.84 | 8.92 |
| d3net-densely-connected-multidilated-densenet | 6.68 | 6.20 | 7.36 | 5.37 | 7.80 |
| lasaft-latent-source-attentive-frequency | 5.88 | 5.63 | 5.68 | 4.87 | 7.33 |
| open-unmix-a-reference-implementation-for | 5.33 | 5.23 | 5.73 | 4.02 | 6.32 |
| music-source-separation-with-band-split-rnn | 8.23 | 7.51 | 8.58 | 6.62 | 10.21 |
| hybrid-transformers-for-music-source | 9.20 | 10.47 | 10.83 | 6.41 | 9.37 |
| contrastive-learning-based-deep-latent | 6.47 | 7.29 | 7.05 | 4.62 | 6.91 |
| d3net-densely-connected-multidilated-densenet | 6.01 | 5.25 | 7.01 | 4.53 | 7.24 |
| all-for-one-and-one-for-all-improving-music | 5.79 | 5.43 | 6.47 | 4.64 | 6.61 |
| hybrid-transformers-for-music-source | 9.00 | 9.78 | 10.08 | 6.42 | 9.20 |
| tf-attention-net-an-end-to-end-neural-network | 5.65 | 5.25 | 6.63 | 4.09 | 6.61 |