# Universal Score-based Speech Enhancement with High Content Preservation

arXiv: https://arxiv.org/abs/2406.12194 · HTML: https://mlsanity.com/paper/2406.12194/

## Key facts

- Cited by 23 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 18 Jun 2024.

## Built on by (most cited first, 23 of 23)

- [High-Resolution Speech Restoration with Latent Diffusion Model](https://mlsanity.com/paper/2409.11145/) (2024)
- [DiTSE: High-Fidelity Generative Speech Enhancement via Latent Diffusion Transformers](https://mlsanity.com/paper/2504.09381/) (2025)
- [Diffusion Buffer: Online Diffusion-based Speech Enhancement with Sub-Second Latency](https://mlsanity.com/paper/2506.02908/) (2025)
- [Vibravox: A Dataset of French Speech Captured with Body-conduction Audio Sensors](https://mlsanity.com/paper/2407.11828/) (2024)
- [Joint Semantic Knowledge Distillation and Masked Acoustic Modeling for Full-band Speech Restoration with Improved Intelligibility](https://mlsanity.com/paper/2409.09357/) (2024)
- [A Composite Predictive-Generative Approach to Monaural Universal Speech Enhancement](https://mlsanity.com/paper/2505.24576/) (2025)
- [VoiceBridge: General Speech Restoration with One-step Latent Bridge Models](https://mlsanity.com/paper/2509.25275/) (2025)
- [EDNet: A Versatile Speech Enhancement Framework with Gating Mamba Mechanism and Phase Shift-Invariant Training](https://mlsanity.com/paper/2506.16231/) (2025)
- [SenSE: Semantic-Aware High-Fidelity Universal Speech Enhancement](https://mlsanity.com/paper/2509.24708/) (2025)
- [Speech Enhancement Based on Drifting Models](https://mlsanity.com/paper/2604.24199/) (2026)
- [CleanMel: Mel-Spectrogram Enhancement for Improving Both Speech Quality and ASR](https://mlsanity.com/paper/2502.20040/) (2025)
- [User-guided Generative Source Separation](https://mlsanity.com/paper/2507.01339/) (2025)
- [Miipher-2: A Universal Speech Restoration Model for Million-Hour Scale Data Restoration](https://mlsanity.com/paper/2505.04457/) (2025)
- [DeepFilterGAN: A Full-band Real-time Speech Enhancement System with GAN-based Stochastic Regeneration](https://mlsanity.com/paper/2505.23515/) (2025)
- [Leveraging Discriminative Latent Representations for Conditioning GAN-Based Speech Enhancement](https://mlsanity.com/paper/2508.20859/) (2025)
- [Geneses: Unified Generative Speech Enhancement and Separation](https://mlsanity.com/paper/2601.18456/) (2026)
- [Quaternion Self-Attention with Shared Scores](https://mlsanity.com/paper/2605.24920/) (2026)
- [A Dual-Branch Parallel Network for Speech Enhancement and Restoration](https://mlsanity.com/paper/2409.08702/) (2024)
- [On the Application of Diffusion Models for Simultaneous Denoising and Dereverberation](https://mlsanity.com/paper/2508.18833/) (2025)
- [Dual-View Predictive Diffusion: Lightweight Speech Enhancement via Spectrogram-Image Synergy](https://mlsanity.com/paper/2602.00568/) (2026)
- [UniPASE: A Generative Model for Universal Speech Enhancement with High Fidelity and Low Hallucinations](https://mlsanity.com/paper/2604.14606/) (2026)
- [RT-SEMamba: Real-Time Speech Enhancement Mamba via Progressive Knowledge Distillation](https://mlsanity.com/paper/2608.12099/) (2026)
- [Cleaner Speech, Weaker Generalization: Revisiting Pitt-Derived Benchmarks for Alzheimer's Disease Detection](https://mlsanity.com/paper/2609.00276/) (2026)

## Abstract (arXiv)

We propose UNIVERSE++, a universal speech enhancement method based on score-based diffusion and adversarial training. Specifically, we improve the existing UNIVERSE model that decouples clean speech feature extraction and diffusion. Our contributions are three-fold. First, we make several modifications to the network architecture, improving training stability and final performance. Second, we introduce an adversarial loss to promote learning high quality speech features. Third, we propose a low-rank adaptation scheme with a phoneme fidelity loss to improve content preservation in the enhanced speech. In the experiments, we train a universal enhancement model on a large scale dataset of speech degraded by noise, reverberation, and various distortions. The results on multiple public benchmark datasets demonstrate that UNIVERSE++ compares favorably to both discriminative and generative baselines for a wide range of qualitative and intelligibility metrics.