2026/03/11 by Duojia Li, Shuhan Zhang, Zihan Qian +5 · 1 voice
Computer Science · Psychology · #Emotion and Mood Recognition #Generalization #Generative model #Matching (statistics) #Pattern recognition (psychology) #Sampling (signal processing) #Similarity (geometry) #Speech Recognition and Synthesis #Speech and Audio Processing #Trajectory #Utterance #cs.AI #cs.SD
paper · pdf · open access · doi:10.48550/arxiv.2603.10701
published in arXiv (Cornell University) (Cornell University)
openalex publication_date 2026/03/11 · arxiv published 2026/03/11 · arxiv updated 2026/03/11 · openalex created_date 2026/03/13 · openalex updated_date 2026/07/28
In target speaker extraction (TSE), we aim to recover target speech from a multi-talker mixture using a short enrollment utterance as reference. Recent studies on diffusion and flow-matching generators have improved target-speech fidelity. However, multi-step sampling increases latency, and one-step solutions often rely on a mixture-dependent time coordinate that can be unreliable for real-world conversations. We present AlphaFlowTSE, a one-step conditional generative model trained with a Jacobian-vector product (JVP)-free AlphaFlow objective. AlphaFlowTSE learns mean-velocity transport along a mixture-to-target trajectory starting from the observed mixture, eliminating auxiliary mixing-ratio prediction, and stabilizes training by combining flow matching with an interval-consistency teacher-student target. Experiments on Libri2Mix and REAL-T confirm that AlphaFlowTSE improves target-speaker similarity and real-mixture generalization for downstream automatic speech recognition (ASR).