Ui-Hyeop Shin

Automatic Speech Recognition (ASR) often degrades in real-world noisy environments, making noise robustness essential for deployment. Supervised noise-augmented fine-tuning is a common remedy, but it can introduce a robustness–clean trade-off and overfit to specific corruptions, degrading recognition in clean conditions. We propose DASH, a self-distillation framework that improves robustness by learning clean–noisy consistency from paired views. DASH distills hidden representations from multiple encoder layers to capture features from low-level acoustics to high-level semantics, and stabilizes training by minimizing KL divergence between prototype assignment distributions of clean and noisy views. Experiments on LibriSpeech show that DASH consistently improves recognition under diverse noisy conditions while preserving clean accuracy, achieved by a label-free pre-training stage with minimal additional overhead (about 4% of fine-tuning time) beyond standard fine-tuning.

This paper presents an efficient speech enhancement (SE) approach that reuses a processing block repeatedly instead of conventional stacking. Rather than increasing the number of blocks for learning deep latent representations, repeating a single block leads to progressive refinement while reducing parameter redundancy. We also minimize domain transformation by keeping encoder and decoder shallow and reusing a single sequence modeling block. Experimental results show that the number of processing stages is more critical to performance than the number of blocks with different weights. Also, we observed that the proposed method gradually refines a noisy input within a single block. Furthermore, with the block reuse method, we demonstrate that deepening the encoder and decoder can be redundant for learning deep complex representation. Therefore, the experimental results confirm that the proposed block reusing enables progressive learning and provides an efficient alternative for SE.

In general, multi-channel source separation has utilized inter-microphone phase differences (IPDs) concatenated with magnitude information in time-frequency domain, or real and imaginary components stacked along the channel axis. However, the spatial information of a sound source is fundamentally contained in the differences between microphones, specifically in the correlation between them, while the power of each microphone also provides valuable information about the source spectrum, which is why the magnitude is also included. Therefore, we propose a network that directly leverages a correlation input with phase transform (PHAT)-beta to estimate the separation filter. In addition, the proposed TF-CorrNet processes the features alternately across time and frequency axes as a dual-path strategy in terms of spatial information. Furthermore, we add a spectral module to model source-related direct time-frequency patterns for improved speech separation. Experimental results demonstrate that the proposed TF-CorrNet effectively separates the speech sounds, showing high performance with a low computational cost in the LibriCSS dataset.

In speaker verification, ECAPA-TDNN has shown remarkable improvement by utilizing one-dimensional(1D) Res2Net block and squeeze-and-excitation(SE) module, along with multi-layer feature aggregation (MFA). Meanwhile, in vision tasks, ConvNet structures have been modernized by referring to Transformer, resulting in improved performance. In this paper, we present an improved block design for TDNN in speaker verification. Inspired by recent ConvNet structures, we replace the SE-Res2Net block in ECAPA-TDNN with a novel 1D two-step multi-scale ConvNeXt block, which we call TS-ConvNeXt. The TS-ConvNeXt block is constructed using two separated sub-modules: a temporal multi-scale convolution (MSC) and a frame-wise feed-forward network (FFN). This two-step design allows for flexible capturing of inter-frame and intra-frame contexts. Additionally, we introduce global response normalization (GRN) for the FFN modules to enable more selective feature propagation, similar to the SE module in ECAPA-TDNN. Experimental results demonstrate that NeXt-TDNN, with a modernized backbone block, significantly improved performance in speaker verification tasks while reducing parameter size and inference time. We have released our code for future studies.

In this paper, we present a statistical beamforming algorithm as a pre-processing step for robust automatic speech recognition (ASR). By modeling the target speech as a non-stationary Laplacian distribution, a mask-based statistical beamforming algorithm is proposed to exploit both its output and masked input variance for robust estimation of the beamformer. In addition, we also present a method for steering vector estimation (SVE) based on a noise power ratio obtained from the target and noise outputs in independent component analysis (ICA). To update the beamformer in the same ICA framework, we derive ICA with distortionless and null constraints on target speech, which yields beamformed speech at the target output and noises at the other outputs, respectively. The demixing weights for the target output result in a statistical beamformer with the weighted spatial covariance matrix (wSCM) using a weighting function characterized by a source model. To enhance the SVE, the strict null constraints imposed by the Lagrange multiplier methods are relaxed by generalized penalties with weight parameters, while the strict distortionless constraints are maintained. Furthermore, we derive an online algorithm based on an optimization technique of recursive least squares (RLS) for practical applications. Experimental results on various environments using CHiME-4 and LibriCSS datasets demonstrate the effectiveness of the presented algorithm compared to conventional beamforming and blind source extraction (BSE) based on ICA on both batch and online processing.

An early conference-stage formulation of statistical beamforming that combines AuxIVA-style demixing updates with distortionless and null constraints on the target speech, aimed at improving robustness for automatic speech recognition. This line of work was later extended into the journal treatment published in IEEE/ACM T-ASLP (2024).

By introducing a frequency dependence source prior including full-band and clique models, independent vector analysis (IVA) has been successfully used for convolutive blind source separation (BSS). In addition, independent low-rank matrix analysis (ILRMA) learns a low-rank approximation of the time-frequency structure of source signals. This paper presents IVA using a clique-based frequency dependence model with time-varying clique variances to combine advantages of both ILRMA and clique-model-based IVA for BSS of speech signals. Although conventional clique models are effective in separating sources with specific spectral structures, the dependency among the cliques is considered by overlaps between cliques or a global clique of all frequency bins if there is. To avoid the permutation problem by strengthening the dependency among the cliques, we develop a generalized probability-density-function (pdf) model imposing a variable exponent on the summed cliques with overlaps and time-varying clique variances, which may include most conventional source models as particular cases. In addition, update rules of the clique variances and demixing matrices are derived by minimization of the cost function of BSS as well as non-negative matrix factorization (NMF) and auxiliary function techniques for fast and robust convergence, respectively. Through experiments on BSS of speech mixtures with various mixing conditions, the proposed IVA showed improved separation performance than the conventional methods. Experimental results consistently demonstrated that the performance of a method could be determined in general by the trade-off between the degree of freedom of source models and the vulnerability to the permutation problem.