Speech Separation On Wsj0 2Mix
المقاييس
Number of parameters (M)
SDRi
SI-SDRi
النتائج
نتائج أداء النماذج المختلفة على هذا المعيار القياسي
جدول المقارنة
| اسم النموذج | Number of parameters (M) | SDRi | SI-SDRi |
|---|---|---|---|
| tf-locoformer-transformer-with-local-modeling | 5.0 | 23 | 22.8 |
| on-time-domain-conformer-models-for-monaural | - | - | 21.2 |
| tasnet-surpassing-ideal-time-frequency | 5.1 | - | 15.3 |
| two-step-sound-source-separation-training-on | - | - | 16.1 |
| tf-locoformer-transformer-with-local-modeling | 15.0 | 23.8 | 23.6 |
| dual-path-transformer-network-direct-context-1 | - | - | 20.2 |
| separate-and-reconstruct-asymmetric-encoder | 59.4 | 25.2 | 25.1 |
| wavesplit-end-to-end-speech-separation-by | - | 22.3 | 22.2 |
| deformable-temporal-convolutional-networks | 3.6 | 17.4 | 17.2 |
| separate-and-diffuse-using-a-pretrained | - | - | 23.9 |
| tasnet-time-domain-audio-separation-network | - | - | 10.8 |
| sandglasset-a-light-multi-granularity-self | - | - | 21.0 |
| spgm-prioritizing-local-features-for-enhanced | 26.2 | - | 22.7 |
| deep-clustering-discriminative-embeddings-for | - | - | 10.8 |
| spgm-prioritizing-local-features-for-enhanced | 26.2 | - | 22.1 |
| compute-and-memory-efficient-universal-sound | - | - | 19.5 |
| boosting-unknown-number-speaker-separation | - | - | 24.0 |
| sepit-approaching-a-single-channel-speech | - | - | 22.4 |
| mossformer-pushing-the-performance-limit-of | 42.1 | - | 22.8 |
| voice-separation-with-an-unknown-number-of | - | - | 20.12 |
| mossformer2-combining-transformer-and-rnn-1 | 55.7 | - | 24.1 |
| alternative-objective-functions-for-deep | - | - | 11.5 |
| sudo-rm-rf-efficient-networks-for-universal | - | - | 18.9 |
| effective-low-cost-time-domain-audio | - | - | 20.3 |
| tf-locoformer-transformer-with-local-modeling | 22.5 | 25.2 | 25.1 |
| attention-is-all-you-need-in-speech | - | 22.4 | 22.3 |
| deformable-temporal-convolutional-networks | 1.3 | 16.3 | 16.1 |
| tf-locoformer-transformer-with-local-modeling | 5.0 | 22.1 | 22 |
| real-time-single-channel-dereverberation-and | - | - | 13.2 |
| divide-and-conquer-a-deep-casa-approach-to | - | - | 17.7 |
| improved-speech-separation-with-time-and | - | - | 16.6 |
| interrupted-and-cascaded-permutation | - | - | 17.5 |
| tf-locoformer-transformer-with-local-modeling | 15.0 | 24.7 | 24.6 |
| self-supervised-pre-training-reduces-label | - | 21.5 | 21.3 |
| mossformer-pushing-the-performance-limit-of | - | - | 22.5 |
| wavesplit-end-to-end-speech-separation-by | - | - | 19.0 |
| dual-path-rnn-efficient-long-sequence | - | - | 18.8 |
| tf-locoformer-transformer-with-local-modeling | 22.5 | 24.3 | 24.2 |