This is the assembled repository of the PASE (Phonologically Anchored Speech Enhancer) family:
-
[2026-06-05] StuPASE: Achieving studio-quality enhancement based on PASE, accepted by Interspeech 2026
StuPASE: Towards Low-Hallucination Studio-Quality Generative Speech Enhancement
-
[2025-11-08] PASE: A low-hallucination SE model, accepted by AAAI 2026
PASE: Leveraging the Phonological Prior of WavLM for Low-Hallucination Generative Speech Enhancement