Selected work

Representative Papers

JSL-DC: A Word-Level Japanese Sign Language Dataset with Linguist-Derived Descriptions for Distinguishing Confusable Signs

Aug 18, 2026

Effective sign language (SL) acquisition is crucial for deaf children, yet 95% are born to hearing parents who often lack proficiency in SL. SL recognition can power learning tools to help parents communicate with their children. However, Japanese Sign Language (JSL) lacks large-scale, multi-signer datasets, hindering the development of models that can generalize to new users. To address this gap, we introduce JSL-DC, the largest JSL dataset by video count, comprising 36.7K videos from 19 signers. The entire process was Deaf-centric: the lexicon comprising 270 JSL words was selected by Deaf and Coda linguists to facilitate parent-child communication, all participants were Deaf individuals who use JSL daily, and the data underwent a two-stage review process involving Deaf linguists. Moreover, we provide linguist-derived descriptions for distinguishing confusable signs. We demonstrate that the proposed model inspired by the descriptions outperforms state-of-the-art recognition methods by 9.8% on the confusable subset. The dataset, along with its linguistic description that inspires new models, will be released under a CC-BY 4.0 license to accelerate research in SL recognition.

0 citationsRead paper

Audio Effect Estimation with DNN-Based Prediction and Search Algorithm

Apr 24, 2026

This study addresses the inverse problem of estimating the types, ordering, and parameter configurations of audio effects applied to a processed (wet) signal. To this end, the authors propose a hybrid strategy that first employs a deep neural network to predict both the dry signal and the combination of effect types, followed by a signal-reconstruction-similarity-driven search algorithm to refine the effect sequence and parameters. This approach represents the first integration of deep neural network prediction with a reconstruction-guided search mechanism, further enhanced by explicit dry signal estimation to improve the accuracy of effect configuration recovery. Experimental results demonstrate that the proposed two-stage hybrid method significantly outperforms purely predictive approaches across multiple evaluation metrics, confirming its effectiveness and superiority in audio effect inversion.

0 citationsRead paper

Recent publications

Latest Papers

JSL-DC: A Word-Level Japanese Sign Language Dataset with Linguist-Derived Descriptions for Distinguishing Confusable Signs

Aug 18, 2026

Effective sign language (SL) acquisition is crucial for deaf children, yet 95% are born to hearing parents who often lack proficiency in SL. SL recognition can power learning tools to help parents communicate with their children. However, Japanese Sign Language (JSL) lacks large-scale, multi-signer datasets, hindering the development of models that can generalize to new users. To address this gap, we introduce JSL-DC, the largest JSL dataset by video count, comprising 36.7K videos from 19 signers. The entire process was Deaf-centric: the lexicon comprising 270 JSL words was selected by Deaf and Coda linguists to facilitate parent-child communication, all participants were Deaf individuals who use JSL daily, and the data underwent a two-stage review process involving Deaf linguists. Moreover, we provide linguist-derived descriptions for distinguishing confusable signs. We demonstrate that the proposed model inspired by the descriptions outperforms state-of-the-art recognition methods by 9.8% on the confusable subset. The dataset, along with its linguistic description that inspires new models, will be released under a CC-BY 4.0 license to accelerate research in SL recognition.

0 citationsRead paper

Audio Effect Estimation with DNN-Based Prediction and Search Algorithm

Apr 24, 2026

This study addresses the inverse problem of estimating the types, ordering, and parameter configurations of audio effects applied to a processed (wet) signal. To this end, the authors propose a hybrid strategy that first employs a deep neural network to predict both the dry signal and the combination of effect types, followed by a signal-reconstruction-similarity-driven search algorithm to refine the effect sequence and parameters. This approach represents the first integration of deep neural network prediction with a reconstruction-guided search mechanism, further enhanced by explicit dry signal estimation to improve the accuracy of effect configuration recovery. Experimental results demonstrate that the proposed two-stage hybrid method significantly outperforms purely predictive approaches across multiple evaluation metrics, confirming its effectiveness and superiority in audio effect inversion.

0 citationsRead paper