GAP-URGENet: A Generative-Predictive Fusion Framework for Universal Speech Enhancement

Xiaobin Rong; Yushi Wang; Zheng Wang; Jing Lu

eess.ASApr 2, 2026

GAP-URGENet: A Generative-Predictive Fusion Framework for Universal Speech Enhancement

Authors: Xiaobin Rong, Yushi Wang, Zheng Wang, Jing Lu

Organizations: Key Laboratory of Modern Acoustics, Nanjing University · NJU-Horizon Intelligent Audio Lab, Horizon Robotics

Abstract

We introduce GAP-URGENet, a generative-predictive fusion framework developed for Track 1 of the ICASSP 2026 URGENT Challenge. The system integrates a generative branch, which performs full-stack speech restoration in a self-supervised representation domain and reconstructs the waveform via a neural vocoder, along with a predictive branch that performs spectrogram-domain enhancement, providing complementary cues. Outputs from both branches are fused by a post-processing module, which also performs bandwidth extension to generate the enhanced waveform at 48 kHz, later downsampled to the original sampling rate. This generative-predictive fusion improves robustness and perceptual quality, achieving top performance in the blind-test phase and ranking 1st in the objective evaluation. Audio examples are available at https://xiaobin-rong.github.io/gap-urgenet_demo.

Figures & tables

Figure 1: Overview of the GAP-URGENet framework.

Model	DNSMOS $\uparrow$	NISQA $\uparrow$	UTMOS $\uparrow$	SCOREQ $\uparrow$	PESQ $\uparrow$	ESTOI $\uparrow$	SBS $\uparrow$	LPS $\uparrow$	SpkSim $\uparrow$
Noisy	1.80	1.56	1.58	1.72	1.26	0.61	0.70	0.58	0.58
BSRNN	3.13	3.48	2.70	3.32	2.57	0.85	0.88	0.82	0.78
BSRNN-Flow	3.19	3.83	2.65	3.35	2.15	0.81	0.83	0.75	0.74
Predictive branch	3.20	3.66	3.13	3.51	2.64	0.86	0.86	0.81	0.70
Generative branch	3.23	4.01	3.01	3.97	2.27	0.83	0.90	0.87	0.76
GAP-URGENet	3.31	3.96	3.22	4.03	2.89	0.90	0.91	0.89	0.82

Table 1: Experimental results on the official validation set.

Speech Processing

Speech processing develops computational methods for recognizing, synthesizing, analyzing, and transforming spoken audio.

See all 412 papers

SB-RF: Schrödinger Bridge Rectified Flow for One-Step Robust Speech Enhancement
Jun 4, 2026Caixia Lu, Xueyang Lv, Penglong Hu +1Rectified Flow Schrödinger Bridges
Predictive-Generative Drift Decomposition for Speech Enhancement and Separation
May 7, 2026Julius Richter, Yoshiki Masuyama, Christoph Boeddeker +3Stochastic Interpolants Speech Processing
MERL Cambridge, MA, USA
DriftSE: Speech Enhancement with Generative Drifting
Sep 10, 2026Liang Xu, Diego Caviedes-Nozal, W. Bastiaan Kleijn +2One-Step Generative Modeling Speech Enhancement
Victoria University of Wellington, New Zealand · GN Advanced Science, Denmark · Lincoln University

Jun 4, 2026J/K move · Enter open · S save

SB-RF: Schrödinger Bridge Rectified Flow for One-Step Robust Speech Enhancement

Caixia Lu, Xueyang Lv, Penglong Hu, Jiaming Xu

Rectified Flow Schrödinger Bridges One-Step Generative Modeling Speech Enhancement Flow-Based Generative Modeling Noise Robustness

Open paper AskPDF arXiv

Generative models have shown promising results for speech enhancement (SE), but they often rely on multi-step inference, limiting low-latency deployment. We propose SB-RF, a one-step generative framework that integrates Rectified Flow (RF) with Schrödinger Bridge (SB) theory. During training, SB-RF samples intermediate states from an SB time marginal and trains a conditional velocity field with the RF velocity-matching objective. At inference, SB-RF starts from the noisy observation and applies a single Euler update. Experiments show that SB-RF achieves competitive performance among generative methods on the VoiceBank-DEMAND benchmark. To further assess performance beyond this standard setting, we evaluate SB-RF on a simulated low signal-to-noise ratio test set using an expanded training dataset. Under these conditions, SB-RF achieves superior performance over the compared baselines, supporting its potential for real-world applications.

Abstract

Speech Processing

Explore similar work

SB-RF: Schrödinger Bridge Rectified Flow for One-Step Robust Speech Enhancement

Predictive-Generative Drift Decomposition for Speech Enhancement and Separation

DriftSE: Speech Enhancement with Generative Drifting