Target Speech Extraction on Relative Cue-based TSE Mixtures (test)
18SI-SDRiTF-Locoformer TSE
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| TF-Locoformer TSEFramework Stage=Two-Stage, Cue Type=Temporal Order2026.03 | 18 | 3.45 | 99.8 | |
| TF-Locoformer TSEFramework Stage=Two-Stage, Cue Type=All Cue2026.03 | 17.4 | 3.41 | 99.8 | |
| TF-Locoformer TSEFramework Stage=Two-Stage, Cue Type=Language2026.03 | 17.3 | 3.41 | 99.3 | |
| TF-Locoformer TSEFramework Stage=Two-Stage, Cue Type=Random Cue2026.03 | 17.1 | 3.39 | 99.2 | |
| TF-Locoformer TSEFramework Stage=Two-Stage, Cue Type=Distance2026.03 | 17 | 3.38 | 98.4 | |
| TF-Locoformer TSEFramework Stage=Two-Stage, Cue Type=Gender2026.03 | 16.7 | 3.37 | 98.1 | |
| TF-Locoformer TSEFramework Stage=Two-Stage, Cue Type=Loudness2026.03 | 16 | 3.35 | 97.2 | |
| TF-Locoformer TSEFramework Stage=Two-Stage, Cue Type=Speaking Duration2026.03 | 15.6 | 3.34 | 95.9 | |
| TF-Locoformer TSEFramework Stage=Two-Stage, Cue Type=Enrollment Audio2026.03 | 15.2 | 3.31 | 95.7 | |
| TF-Locoformer TSEFramework Stage=Single-Stage, Cue Type=Temporal Order2026.03 | 15.1 | 2.99 | — | |
| TF-Locoformer TSEFramework Stage=Two-Stage, Cue Type=Pitch Level2026.03 | 14.8 | 3.3 | 94.6 | |
| TF-Locoformer TSEFramework Stage=Single-Stage, Cue Type=Distance2026.03 | 14.6 | 2.89 | — | |
| TF-Locoformer TSEFramework Stage=Single-Stage, Cue Type=All Cue2026.03 | 14.3 | 2.88 | — | |
| TF-Locoformer TSEFramework Stage=Single-Stage, Cue Type=Loudness2026.03 | 13.7 | 2.84 | — | |
| TF-Locoformer TSEFramework Stage=Single-Stage, Cue Type=Random Cue2026.03 | 13.2 | 2.7 | — | |
| TF-Locoformer TSEFramework Stage=Two-Stage, Cue Type=Speaking Rate2026.03 | 13.1 | 3.27 | 91.3 | |
| TF-Locoformer TSEFramework Stage=Single-Stage, Cue Type=Speaking Duration2026.03 | 13 | 2.8 | — | |
| TF-Locoformer TSEFramework Stage=Single-Stage, Cue Type=Gender2026.03 | 12.8 | 2.75 | — | |
| TF-Locoformer TSEFramework Stage=Single-Stage, Cue Type=Pitch Level2026.03 | 12.8 | 2.75 | — | |
| TF-Locoformer TSEFramework Stage=Two-Stage, Cue Type=Transcription2026.03 | 12.5 | 3.22 | 90.4 | |
| TF-Locoformer TSEFramework Stage=Single-Stage, Cue Type=Language2026.03 | 11.4 | 2.67 | — | |
| TF-Locoformer TSEFramework Stage=Two-Stage, Cue Type=Pitch Range2026.03 | 11.3 | 3.14 | 88.7 | |
| TF-Locoformer TSEFramework Stage=Two-Stage, Cue Type=Emotion2026.03 | 10.3 | 3.1 | 87.2 | |
| TF-Locoformer TSEFramework Stage=Single-Stage, Cue Type=Pitch Range2026.03 | 9 | 2.47 | — | |
| DPT-based TSEFramework Stage=Single-Stage, Cue Type=All Cue2026.03 | 8.9 | 1.8 | — | |
| DPT-based TSEFramework Stage=Single-Stage, Cue Type=Gender2026.03 | 8.8 | 1.8 | — | |
| TF-Locoformer TSEFramework Stage=Single-Stage, Cue Type=Transcription2026.03 | 8.7 | 2.47 | — | |
| DPT-based TSEFramework Stage=Single-Stage, Cue Type=Loudness2026.03 | 8.7 | 1.8 | — | |
| DPT-based TSEFramework Stage=Single-Stage, Cue Type=Temporal Order2026.03 | 8.7 | 1.85 | — | |
| DPT-based TSEFramework Stage=Single-Stage, Cue Type=Random Cue2026.03 | 8.4 | 1.78 | — | |
| DPT-based TSEFramework Stage=Single-Stage, Cue Type=Distance2026.03 | 8.3 | 1.78 | — | |
| DPT-based TSEFramework Stage=Single-Stage, Cue Type=Pitch Level2026.03 | 7.9 | 1.77 | — | |
| TF-Locoformer TSEFramework Stage=Single-Stage, Cue Type=Speaking Rate2026.03 | 7.9 | 2.44 | — | |
| TF-Locoformer TSEFramework Stage=Single-Stage, Cue Type=Emotion2026.03 | 7.8 | 2.44 | — | |
| DPT-based TSEFramework Stage=Single-Stage, Cue Type=Speaking Duration2026.03 | 7.4 | 1.79 | — | |
| DPT-based TSEFramework Stage=Single-Stage, Cue Type=Language2026.03 | 5.5 | 1.71 | — | |
| DPT-based TSEFramework Stage=Single-Stage, Cue Type=Pitch Range2026.03 | 5.3 | 1.68 | — | |
| DPT-based TSEFramework Stage=Single-Stage, Cue Type=Emotion2026.03 | 4.9 | 1.56 | — | |
| DPT-based TSEFramework Stage=Single-Stage, Cue Type=Transcription2026.03 | 4.3 | 1.67 | — | |
| DPT-based TSEFramework Stage=Single-Stage, Cue Type=Speaking Rate2026.03 | 4 | 1.68 | — | |
| DPT-based TSEFramework Stage=Single-Stage, Cue Type=Age Category2026.03 | -1 | 1.38 | — | |
| TF-Locoformer TSEFramework Stage=Two-Stage, Cue Type=Age Category2026.03 | -1.6 | 2.45 | 64.3 | |
| TF-Locoformer TSEFramework Stage=Single-Stage, Cue Type=Age Category2026.03 | -1.9 | 1.71 | — |