研究者詳細 - 水町　光徳

論文 - 水町　光徳

分割表示195 件中 21 - 40 件目／全件表示 >>

Subband Optimization of Nonlinear Beamformer with Long Short-Term Memory in Low-Frequency Region 査読有り国際誌

M. Mizumachi, K. Nashiro

Proc. 29th International Congress on Sound and Vibration (ICSV29) Paper ID: 483 2023年07月

音メディアを用いた実験を支援するための対話的測定ツールについて

河原英紀, 矢田部浩平, 榊原健一, 水町光徳, 程島奈緒, 松井淑恵

音楽音響研究会資料 42 ( 3 ) 41 - 44 2023年07月

Singing Voice Range Profiling Toolbox with real-time interaction and its application to make recording data reusable 査読有り国際誌

H. Kawahara, KI. Sakakibara, K. Yatabe, M. Mizumachi, T. Kitamura

Proc. Stockholm Music Acoustics Conference 2023 (SMAC 2023) 160 - 166 2023年06月

Relationship between perceptual width of stereo stage and inter-channel cross-correlation 査読有り国際誌

M. Mizumachi, T. Horiuchi

Proc. 154th AES Convention Express Paper 68 2023年05月

Analysis of Temporal Changes in Prosodic Features of Positive and Negative Emotions in Spontaneous Speech 査読有り

Y. Aoki, M. Mizumachi

Proc. ICIAE2023 VS-2 2023年03月

Relationship between the Temporal Variation of Fundamental Frequency and Subjective Hoarseness 査読有り

S. Tomimatsu, M. Mizumachi

Proc. ICIAE2023 PS-4 2023年03月

SUBBAND OPTIMIZATION OF NONLINEAR BEAMFORMER WITH LONG SHORT-TERM MEMORY IN LOW-FREQUENCY REGION 査読有り国際誌

Mizumachi M., Nashiro K.

Proceedings of the International Congress on Sound and Vibration 2023年01月

担当区分：筆頭著者,　責任著者記述言語：英語掲載種別：研究論文（国際会議プロシーディングス）

Acoustic beamforming is essential to achieve signal enhancement, noise reduction, sound source localization, and acoustic imaging in various applications. Traditional beamformers, such as delay-and-sum beamformers, have been designed by linear signal processing and diversified in adaptive filtering, statistical signal processing, and machine learning. Nonlinear beamformers optimized by neural networks have great potential to optimize their beampatterns. The nonlinear beamformer trained with a single sinusoidal signal succeeds in sharpening its beam-pattern at the specific frequency. The authors have proposed a wideband nonlinear beamformer to sharpen the mainlobe in the speech band. It was, however, challenging to achieve a sharp mainlobe in a lower frequency range below 500 Hz with an 8-ch small-scale microphone array. This paper investigates sub-band optimization of the nonlinear beamformer in the lower frequencies. It is confirmed that the nonlinear beamformer with long short-term memory, suitable cell size, and unit number could sharpen the mainlobe without sidelobes in the lower frequency range.

Scopus

その他リンク： https://www.scopus.com/inward/record.uri?partnerID=HzOxMe3b&scp=85170649321&origin=inward

Relationship between perceptual width of stereo stage and inter-channel cross-correlation 査読有り

Mizumachi M., Horiuchi T.

AES Europe 2023: 154th Audio Engineering Society Convention 2023年01月

担当区分：筆頭著者,　責任著者記述言語：英語掲載種別：研究論文（国際会議プロシーディングス）

It is desirable to properly generate inter-channel crosstalk while reproducing legacy stereo sound sources through earphones and headphones. The authors proposed amplitude-based and phase-based crosstalk generators, of which parameters were perceptually optimized by performing a listening test. The optimal parameters were determined based on the perceptual width of the stereo stage by comparing the proposed method with binaural rendering. This paper investigates the relationship between the perceptual stereo stage width and inter-channel cross-correlation. It is confirmed that the inter-channel cross-correlation has non-linear relation to the perceptual stereo stage width. This result contributes to efficiently optimizing the crosstalk generators without subject evaluation.

Scopus

その他リンク： https://www.scopus.com/inward/record.uri?partnerID=HzOxMe3b&scp=85174508475&origin=inward

Higher accuracy self-supervised visual odometry with reliable projection 査読有り

Zhou S., Yang Z., Zhu M., Li H., Serikawa S., Mizumachi M., Zhang L.

Artificial Life and Robotics 27 ( 3 ) 568 - 575 2022年08月

記述言語：英語掲載種別：研究論文（学術雑誌）

A complex outdoor environment includes variable illuminations, dynamic objects, and occlusion, making visual odometry in outdoor environments remains a challenging problem. This work trains a self-supervised pose estimation network based on the projection between adjacent frames. The pose between two adjacent frames is a 6D vector. It can be better solved with reliable pixels, instead of all the pixels. Thus, only reliable pixels are under consideration when training. This research proposed a boundary mask and inferior projection mask to eliminate the influence of unreliable pixels. Furthermore, we evaluated the proposed method on the KITTI datasets and compared it with other state-of-the-art deep learning-based visual odometry methods. The result shows that the proposed method can detect inferior projection and boundary pixels well. Moreover, it achieves higher accuracy in pose estimation.

DOI： 10.1007/s10015-022-00766-7

Scopus

その他リンク： https://www.scopus.com/inward/record.uri?partnerID=HzOxMe3b&scp=85131664759&origin=inward

Optimization of Nonlinear Beamformer with Long Short-term Memory 査読有り

M. Mizumachi, S. Arase

Proc. 28th International Congress on Sound and Vibration (ICSV28) Paper ID: 707 2022年07月

Perceptual optimization of hybrid stereo width control method compared with loudspeakers reproduction 査読有り国際誌

M. Mizumachi, Y. Ueno, T. Horiuchi

Proc. 152nd AES Convention Paper Number: 10571 2022年05月

Perceptual optimization of stereo width shrinkage method in headphone reproduction 査読有り

Ueno Y., Mizumachi M., Horiuchi T.

Acoustical Science and Technology ( 一般社団法人日本音響学会 ) 43 ( 2 ) 170 - 172 2022年03月

特集―高臨場感オーディオ―誌上座談会

安藤彰男, 小澤賢司, 亀川徹, 中原雅考, 西口正之, 青木秀一, 杉本岳大, 水町光徳

日本音響学会誌 ( 一般社団法人日本音響学会 ) 78 ( 3 ) 108 - 113 2022年03月

Inter-channel Cross-correlations of Stereo Width Shrinkage Methods Based on Amplitude and Phase Differences 査読有り

Y. Ueno, M. Mizumachi, T. Horiuchi

Proc. ICIAE2022 VS2-2 2022年03月

Investigation of Suitable Training Conditions for Nonlinear Beamformer 査読有り

S. Arase, M. Mizumachi

Proc. ICIAE2022 PS-2 2022年03月

音声収録におけるスマートデバイスなどを用いた音圧較正の検討

河原英紀, 榊原健一, 水町光徳, 北村達也

音声コミュニケーション研究会資料 ( 音声コミュニケーション研究委員会 ) 2 ( 1 ) n/a 2022年01月

記述言語：日本語掲載種別：研究論文（研究会，シンポジウム資料等）

<p>音声研究のための利用価値の高い音声素材の収録のためには、収録の物理的条件を記録する必要がある。録音系の音圧較正は、そのための出発点となる重要なステップである。ここでは、広く普及しているスマートデバイスを用いた較正の可能性についての検討状況について報告する。日常の環境において適切な較正を行うためには、機器そのものとあわせて、実施可能な手順と、それらに伴う誤差を把握しておく必要がある。ここでは、デジタルデバイスと離散時間系の信号処理の特徴を活かした、幾つかの方法について紹介する。</p>

DOI： 10.60274/asjsc.sc-2022-1

CiNii Research

OPTIMIZATION OF NONLINEAR BEAMFORMER WITH LONG SHORT-TERM MEMORY 査読有り

Mizumachi M., Arase S.

Proceedings of the International Congress on Sound and Vibration 2022年01月

担当区分：筆頭著者,　責任著者記述言語：英語掲載種別：研究論文（国際会議プロシーディングス）

Nowadays, increasing demand for teleworking needs audio capturing systems that can suppress noises and emphasize the target speech. Beamforming, which achieves spatial filtering using a microphone array, is one of the widely-used techniques for speech enhancement and noise suppression. The state-of-the-art nonlinear beamformers are designed with neural networks. In this study, the neural network with the Long Short-Term Memory as a recursive neural network is employed for dealing with acoustic signals as time series. In this study, the proposed beamformer using the neural network is optimized based on the frequency characteristics of the training data and the results of objective evaluation. The training data were prepared using an 8-ch small-scale microphone array with 10 mm microphone spacing in a soundproofed room. The experimental results show that the training data, which consist of the mixture of 1 kHz and 4 kHz sinusoidal signals, achieve suppressing noises coming from non-target directions and enhancing the target signal with low distortion.

Scopus

その他リンク： https://www.scopus.com/inward/record.uri?partnerID=HzOxMe3b&scp=85149853200&origin=inward

A Method to Automatic Create Dataset for Training Object Detection Neural Networks 査読有り

Zhou S., Yang Z., Zhu M., Li H., Serikawa S., Mizumachi M., Zhang L.

IEEE Access 10 80505 - 80517 2022年01月

記述言語：英語掲載種別：研究論文（学術雑誌）

Numerous high-accuracy neural networks for object detection have been proposed in recent years. Creating a specific training dataset seems to be the main obstacle in putting them into practice. This paper aims to overcome this obstacle and proposes an automatic dataset creation method. This method first extracts objects from the source images and then combines them as synthetic images. These synthetic images are annotated automatically using the data flow in the process and can be used directly for training. To the best of our knowledge, this is the first automatic dataset creation method for object detection tasks. In addition, the adaptive object extraction method and created natural synthetic images make the proposed method maintain strong adaptation and generalization ability. To validate the feasibility, a dataset that includes 44 categories of objects is created for the object detection task in a vending supermarket. Under the strictest metric AP75, both the trained EfficientDet and YOLOv4 achieve higher than 95% in accuracy on the common difficulty testing set and higher than 90% in accuracy on the high difficulty testing set.

DOI： 10.1109/ACCESS.2022.3195490

Scopus

その他リンク： https://www.scopus.com/inward/record.uri?partnerID=HzOxMe3b&scp=85135766558&origin=inward

Perceptual optimization of hybrid stereo width control method compared with loudspeakers reproduction 査読有り

Mizumachi M., Ueno Y., Horiuchi T.

AES Europe Spring 2022 - 152nd Audio Engineering Society Convention 2022 236 - 240 2022年01月

担当区分：筆頭著者,　責任著者記述言語：英語掲載種別：研究論文（国際会議プロシーディングス）

Legacy stereo music sources cause unnatural spatial impressions through the earphones and headphones reproduction due to the lack of crosstalk, which is naturally produced in loudspeakers reproduction. The amplitude-based and phase-based hybrid stereo width control method is proposed to generate the crosstalk components between channels in headphones reproduction. The control parameters for both the previously-proposed amplitude-based and hybrid methods are perceptually optimized compared with the loudspeaker reproduction. Suitable parameter values are successfully obtained for both methods with classical music sources. It is confirmed that the suitable parameter values have a different trend in those methods depending on the music sources.

Scopus

その他リンク： https://www.scopus.com/inward/record.uri?partnerID=HzOxMe3b&scp=85136323424&origin=inward

公共空間の音声明瞭度を改善するEGG解析に基づく話者指標の検討

程島奈緒, 榊原健一, 水町光徳

日本音響学会春季研究発表会講演論文集 - 1123 - 1124 2022年01月

　前のページ - 次のページ　

<水町 光徳>

論文 - 水町 光徳

<水町　光徳>

論文 - 水町　光徳