SigWavNet: Learning Multiresolution Signal Wavelet Network for Speech Emotion Recognition

2025-02-01

AI Summary

To address poor robustness, weak discriminability, and noise sensitivity in raw-waveform modeling for speech emotion recognition (SER), this paper proposes an end-to-end multi-resolution waveform modeling framework. Methodologically, it introduces: (1) learnable wavelet bases coupled with adaptive hard-thresholding denoising for localized time-frequency noise suppression; (2) a CNN-GRU architecture integrating spatio-temporal dual attention, 1D dilated convolutions, and bidirectional gated recurrent units to enhance long-range temporal dependency modeling; and (3) native support for variable-length utterances via direct end-to-end processing—eliminating segmentation and handcrafted feature engineering. Evaluated on IEMOCAP and EMO-DB, the framework achieves significant improvements over state-of-the-art methods, demonstrating superior noise robustness, emotional discriminability, and architectural simplicity.

Technology Category

• Natural Language Processing: Speech
• Machine Learning: Multimodal Learning
• Cognitive Modeling & Cognitive Systems: Neural Spike Coding

Application Category

• Search and Retrieval-Augmented AI: Multilingual and cross-lingual Web search
• Graph Algorithms and Modeling for the Web: Graph embeddings and representation learning for Web-related graphs
• Web Mining and Content Analysis: Mining multimedia, multimodal, multilingual, cross-lingual Web data

Abstract

In the field of human-computer interaction and psychological assessment, speech emotion recognition (SER) plays an important role in deciphering emotional states from speech signals. Despite advancements, challenges persist due to system complexity, feature distinctiveness issues, and noise interference. This paper introduces a new end-to-end (E2E) deep learning multi-resolution framework for SER, addressing these limitations by extracting meaningful representations directly from raw waveform speech signals. By leveraging the properties of the fast discrete wavelet transform (FDWT), including the cascade algorithm, conjugate quadrature filter, and coefficient denoising, our approach introduces a learnable model for both wavelet bases and denoising through deep learning techniques. The framework incorporates an activation function for learnable asymmetric hard thresholding of wavelet coefficients. Our approach exploits the capabilities of wavelets for effective localization in both time and frequency domains. We then combine one-dimensional dilated convolutional neural networks (1D dilated CNN) with a spatial attention layer and bidirectional gated recurrent units (Bi-GRU) with a temporal attention layer to efficiently capture the nuanced spatial and temporal characteristics of emotional features. By handling variable-length speech without segmentation and eliminating the need for pre or post-processing, the proposed model outperformed state-of-the-art methods on IEMOCAP and EMO-DB datasets. The source code of this paper is shared on the Github repository: https://github.com/alaaNfissi/SigWavNet-Learning-Multiresolution-Signal-Wavelet-Network-for-Speech-Emotion-Recognition.

Problem

• Speech Emotion Recognition
• Acoustic Complexity
• Noise Robustness

Innovation

• SigWavNet
• FDWT
• Emotion Recognition