Determined blind source separation via modeling adjacent frequency band correlations in speech signals

2025-04-05

AI Summary

Conventional frequency-domain blind source separation (BSS) methods—such as Independent Low-Rank Matrix Analysis (ILRMA) and Multichannel Nonnegative Matrix Factorization (MNMF), which assume inter-frequency bin independence, or Independent Vector Analysis (IVA), which assumes uniform correlation across frequencies—fail to accurately model the strong local spectral correlation between adjacent frequency bins in speech signals. Method: This paper proposes a novel paradigm for modeling local spectral dependencies by introducing, for the first time, a prior on adjacent-band correlation, embedded within an enhanced weighted Sinkhorn ILRMA (wsILRMA) framework. It employs the weighted Sinkhorn divergence to optimize the joint probability distribution of source spectra. Contribution/Results: By explicitly encoding the physically grounded high correlation among neighboring frequency bins, wsILRMA transcends traditional statistical assumptions. Evaluated on standard benchmarks, it achieves significant improvements in Signal-to-Distortion Ratio (SDR) and Signal-to-Interference Ratio (SIR), consistently outperforming ILRMA, MNMF, and IVA.

Technology Category

• Machine Learning: Semi-Supervised Learning
• Natural Language Processing: Speech
• Cognitive Modeling & Cognitive Systems: Neural Spike Coding

Application Category

• Web Mining and Content Analysis: Robustness and generalizability of Web mining methods
• Search and Retrieval-Augmented AI: Multilingual and cross-lingual Web search
• User Modeling, Personalization and Recommendation: User modeling and simulation for interactive and conversational systems

Abstract

Multichannel blind source separation (MBSS), which focuses on separating signals of interest from mixed observations, has been extensively studied in acoustic and speech processing. Existing MBSS algorithms, such as independent low-rank matrix analysis (ILRMA) and multichannel nonnegative matrix factorization (MNMF), utilize the low-rank structure of source models but assume that frequency bins are independent. In contrast, independent vector analysis (IVA) does not rely on a low-rank source model but rather captures frequency dependencies based on a uniform correlation assumption. In this work, we demonstrate that dependencies between adjacent frequency bins are significantly stronger than those between bins that are farther apart in typical speech signals. To address this, we introduce a weighted Sinkhorn divergence-based ILRMA (wsILRMA) that simultaneously captures these inter-frequency dependencies and models joint probability distributions. Our approach incorporates an inter-frequency correlation constraint, leading to improved source separation performance compared to existing methods, as evidenced by higher Signal-to-Distortion Ratios (SDRs) and Source-to-Interference Ratios (SIRs).

Problem

• Enhances blind source separation by modeling adjacent frequency correlations
• Improves speech signal separation using weighted Sinkhorn divergence-based ILRMA
• Addresses frequency dependency limitations in existing MBSS algorithms

Innovation

• Weighted Sinkhorn divergence-based ILRMA
• Models adjacent frequency band correlations
• Improves speech signal separation performance