Hold Me Tight: Stable Encoder-Decoder Design for Speech Enhancement

Autor:	Haider, Daniel, Perfler, Felix, Lostanlen, Vincent, Ehler, Martin, Balazs, Peter
Rok vydání:	2024
Předmět:	Computer Science - Sound Computer Science - Machine Learning Electrical Engineering and Systems Science - Audio and Speech Processing
Druh dokumentu:	Working Paper
Popis:	Convolutional layers with 1-D filters are often used as frontend to encode audio signals. Unlike fixed time-frequency representations, they can adapt to the local characteristics of input data. However, 1-D filters on raw audio are hard to train and often suffer from instabilities. In this paper, we address these problems with hybrid solutions, i.e., combining theory-driven and data-driven approaches. First, we preprocess the audio signals via a auditory filterbank, guaranteeing good frequency localization for the learned encoder. Second, we use results from frame theory to define an unsupervised learning objective that encourages energy conservation and perfect reconstruction. Third, we adapt mixed compressed spectral norms as learning objectives to the encoder coefficients. Using these solutions in a low-complexity encoder-mask-decoder model significantly improves the perceptual evaluation of speech quality (PESQ) in speech enhancement. Comment: Accepted at INTERSPEECH 2024
Databáze:	arXiv
Externí odkaz:	http://arxiv.org/abs/2408.17358 Zobrazit plný text záznamu View this record from Arxiv