# Mitigating Over-Suppression in Speech Enhancement via Inference-Time Rethink-and-Refine Correction Module

arXiv: https://arxiv.org/abs/2608.07781 · HTML: https://mlsanity.com/paper/2608.07781/

## Key facts

- Cited by 1 papers in the map (data through 19 Oct 2026).
- Placed in Speech enhancement and noise suppression, one of 11 directions in Speech and audio language models.
- Builds on 19 of its 29 references that are in the map.
- First posted on arXiv 7 Aug 2026.

## Placed in

- Speech and audio language models / [Speech enhancement and noise suppression](https://mlsanity.com/speech/speech-enhancement-and-noise-suppression/)

## Builds on (most cited first, 19 of 19)

- [Robust Speech Recognition via Large-Scale Weak Supervision](https://mlsanity.com/paper/2212.04356/) (2022)
- [Reflexion: Language Agents with Verbal Reinforcement Learning](https://mlsanity.com/paper/2303.11366/) (2023)
- [Self-Refine: Iterative Refinement with Self-Feedback](https://mlsanity.com/paper/2303.17651/) (2023)
- [Speech Enhancement and Dereverberation with Diffusion-based Generative Models](https://mlsanity.com/paper/2208.05830/) (2022)
- [Meta Audiobox Aesthetics: Unified Automatic Quality Assessment for Speech, Music, and Sound](https://mlsanity.com/paper/2502.05139/) (2025)
- [An Investigation of Incorporating Mamba for Speech Enhancement](https://mlsanity.com/paper/2405.06573/) (2024)
- [MP-SENet: A Speech Enhancement Model with Parallel Denoising of Magnitude and Phase Spectra](https://mlsanity.com/paper/2305.13686/) (2023)
- [CMGAN: Conformer-based Metric GAN for Speech Enhancement](https://mlsanity.com/paper/2203.15149/) (2022)
- [How Bad Are Artifacts?: Analyzing the Impact of Speech Enhancement Errors on ASR](https://mlsanity.com/paper/2201.06685/) (2022)
- [SCOREQ: Speech Quality Assessment with Contrastive Regression](https://mlsanity.com/paper/2410.06675/) (2024)
- [URGENT Challenge: Universality, Robustness, and Generalizability For Speech Enhancement](https://mlsanity.com/paper/2406.04660/) (2024)
- [Dense CNN with Self-Attention for Time-Domain Speech Enhancement](https://mlsanity.com/paper/2009.01941/) (2020)
- [Phone-to-audio alignment without text: A Semi-supervised Approach](https://mlsanity.com/paper/2110.03876/) (2021)
- [Interspeech 2025 URGENT Speech Enhancement Challenge](https://mlsanity.com/paper/2505.23212/) (2025)
- [SpeakerNet: 1D Depth-wise Separable Convolutional Network for Text-Independent Speaker Recognition and Verification](https://mlsanity.com/paper/2010.12653/) (2020)
- [InQSS: a speech intelligibility and quality assessment model using a multi-task learning network](https://mlsanity.com/paper/2111.02585/) (2021)
- [Noise robust speech emotion recognition with signal-to-noise ratio adapting speech enhancement](https://mlsanity.com/paper/2309.01164/) (2023)
- [Multi-CMGAN+/+: Leveraging Multi-Objective Speech Quality Metric Prediction for Speech Enhancement](https://mlsanity.com/paper/2312.08979/) (2023)
- [Too Good to Be True: A Study on Modern Automatic Speech Recognition for the Evaluation of Speech Enhancement](https://mlsanity.com/paper/2605.12107/) (2026)

## Built on by (most cited first, 1 of 1)

- [BLINC: Blind Calibration For Training-Free Speech Enhancement Adaptation](https://mlsanity.com/paper/2609.21898/) (2026)

## Abstract (arXiv)

We present a rethink-and-refine correction module that addresses over-suppression, a common failure mode of speech enhancement (SE) models, where speech cues are suppressed alongside noise. Our method operates entirely in the inference stage without additional training, allowing seamless integration with diverse SE models. Given noisy and enhanced signals, we obtain word- or phoneme-level alignments using an automatic speech recognition model and identify intervals where enhancement is unreliable. These intervals are then selectively remixed through convex interpolation, with per-segment weights optimized to maximize a composite objective balancing perceptual quality and speech preservation. Experiments on the URGENT 2024 and 2025, VCTK-DEMAND, and MSP-PODCAST datasets show consistent improvements in perceptual quality, intelligibility, and downstream performance compared to conventional SE alone, demonstrating the benefit of rethink-and-refine framework for robust speech processing.