Abstract
확산 기반 발화 개선 시스템의 소개
제안된 통합 시스템의 구조와 작동 방식
실험 결과 및 성능 평가
다른 방법과의 비교
Stats
"Experiments conducted on the Voice-Bank dataset demonstrate that incorporating predictive information leads to faster decoding and higher PESQ scores compared with other score-based diffusion SE (StoRM and SGMSE+)."
"The performance improves as the iteration steps increases."
"The proposed system can combine the characteristics of predictive and generative SE and fully use the predictive complex spectrograms even when the number of diffusion steps is small."
Quotes
"Diffusion-based generative speech enhancement (SE) has recently received attention, but reverse diffusion remains time-consuming."
"In this paper, we propose a unified system that use jointly generative and predictive decoders across two levels."
"The predictive information helps the model to reduce speech distortion, noise, and artifacts."