mlsanityData to 7 Oct 2026

VorTEX: Various overlap ratio for Target speech EXtraction

arXiv 2603.14803 ↗First posted 16 Mar 2026cs.SD cs.AI cs.CLcited by 0

Not on the map. Why?

Abstract (arXiv)

Target speech extraction (TSE) aims to recover a target speaker's voice from a mixture. While recent text-prompted approaches have shown promise, most approaches assume fully overlapped mixtures, limiting insight into behavior across realistic overlap ratios. We introduce VorTEX (Various overlap ratio for Target speech EXtraction), a text-prompted TSE architecture with a Decoupled Adaptive Multi-branch (DAM) Fusion block that separates primary extraction from auxiliary regularization pathways. To enable controlled analysis, we construct PORTE, a two-speaker dataset spanning overlap ratios from 0% to 100%. We further propose Suppression Ratio on Energy (SuRE), a diagnostic metric that detects suppression behavior not captured by conventional measures. Experiments show that existing models exhibit suppression or residual interference under overlap, whereas VorTEX achieves the highest separation fidelity across 20-100% overlap (e.g., 5.50 dB at 20% and 2.04 dB at 100%) while maintaining zero SuRE, indicating robust extraction without suppression-driven artifacts.

Key facts. No citing papers in the map yet (data through 7 Oct 2026). Builds on 23 of its 28 references that are in the map. First posted on arXiv 16 Mar 2026.

Builds on

23 of 28 references.

References · 23

  1. 2017Attention Is All You Needcited by 59,932
  2. 2021RoFormer: Enhanced Transformer with Rotary Position Embeddingcited by 4,581
  3. 2022Robust Speech Recognition via Large-Scale Weak Supervisioncited by 3,972
  4. 2022Scaling Instruction-Finetuned Language Modelscited by 2,997
  5. 2019LibriTTS: A Corpus Derived from LibriSpeech for Text-to-Speechcited by 639
  6. 2018SDR - half-baked or well done?cited by 420
  7. 2018Conv-TasNet: Surpassing Ideal Time-Frequency Magnitude Masking for Speech Separationcited by 400
  8. 2015Deep clustering: Discriminative embeddings for segmentation and separationcited by 211
  9. 2020LibriMix: An Open-Source Dataset for Generalizable Speech Separationcited by 154
  10. 2020Continuous speech separation: dataset and analysiscited by 73
  11. 2024Restructuring Vector Quantization with the Rotation Trickcited by 49
  12. 2023TextrolSpeech: A Text Style Control Speech Corpus With Codec Language Text-to-Speech Modelscited by 48
11 more
  1. 2023Separate Anything You Describecited by 43
  2. 2023Typing to Listen at the Cocktail Party: Text-Guided Target Speaker Extractioncited by 6
  3. 2024SpeakerBeam-SS: Real-time Target Speaker Extraction with Lightweight Conv-TasNet and State Space Modelingcited by 5
  4. 2021SDR -- Medium Rare with Fast Computationscited by 4
  5. 2023USED: Universal Speaker Extraction and Diarizationcited by 4
  6. 2025A Study of the Scale Invariant Signal to Distortion Ratio in Speech Separation with Noisy Referencescited by 4
  7. 2025Beyond Speaker Identity: Text Guided Target Speech Extractioncited by 3
  8. 2024Continuous Target Speech Extraction: Enhancing Personalized Diarization and Extraction on Complex Recordingscited by 2
  9. 2025Universal Speaker Embedding Free Target Speaker Extraction and Personal Voice Activity Detectioncited by 2
  10. 2025DGMO: Training-Free Audio Source Separation through Diffusion-Guided Mask Optimizationcited by 2
  11. 2025Inter-Speaker Relative Cues for Text-Guided Target Speech Extractioncited by 1

Built on by

Not cited yet.