研究者詳細 - 水町　光徳

論文 - 水町　光徳

分割表示195 件中 81 - 100 件目／全件表示 >>

Power Saving Audio Playback Algorithm Based on Auditory Characteristics 査読有り

M. Mizumachi, W. Kubota, M. Nakagawara

Proc. AES 144th Convention Paper No. 9933 2018年05月

Noise reduction using neural network trained with amplitude spectra 査読有り

Mizumachi M., Otani R.

INTER-NOISE 2018 - 47th International Congress and Exposition on Noise Control Engineering: Impact of Noise Control Engineering 2018年01月

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

© INTER-NOISE 2018 - 47th International Congress and Exposition on Noise Control Engineering: Impact of Noise Control Engineering. All rights reserved. Noise reduction has been one of the most important issues in acoustic signal processing. There are a wide variety of techniques to enhance the target signal and reduce interference. In recent days, neural network-based approaches achieve a great noise reduction performance. The authors have also proposed a neural network-based beamformer, that is, an autoencoder in the temporal domain. Channel-dependent waveforms obtained by an 8-ch linear microphone array were prepared as training data. It succeeds in forming a sharp main-lobe and suppressing side-lobes when the interference comes from the direction away from the target direction. Its performance drops when the direction of the interference is near that of the target signal. In this paper, the neural network-based beamformer is trained with amplitude spectra of the harmonic signals such as musical instruments. In other words, phase information is not used for noise reduction. The performance of the proposed method is evaluated based on signal-to-noise ratio and spectral distortion. The proposed method has an advantage in noise reduction for interferences, of which arrival directions are close to the target direction.

Scopus

その他リンク： https://www.scopus.com/inward/record.uri?partnerID=HzOxMe3b&scp=85059360819&origin=inward

What is speech intelligibility? 査読有り

Mizumachi M.

25th International Congress on Sound and Vibration 2018, ICSV 2018: Hiroshima Calling 6 3633 - 3636 2018年01月

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

Copyright © (2018) by International Institute of Acoustics & Vibration. All rights reserved. Speech intelligibility is an important factor but is partly considered for each purpose. The intelligibility of clean speech is difficult to be quantified, although the speech intelligibility indexes in noisy and reverberant environments are widely standardized. In this study, the relationship between the vowel space and subjective impressions is investigated for Japanese five vowels. The vowel space, that is the pentagon defined by the first and second formant frequencies, is considered in the Equivalent Rectangular Bandwidth scale, which is an auditory-oriented, non-linear frequency scale. Concerning the subjective evaluation, a listening test is carried out with five grade scale on subjective fluency and so on. It is found that the skewness of the vowel space is highly correlated with the subjective fluency of vowels.

Scopus

その他リンク： https://www.scopus.com/inward/record.uri?partnerID=HzOxMe3b&scp=85058781649&origin=inward

Power saving audio playback algorithm based on auditory characteristics 査読有り

Mizumachi M., Kubota W., Nakagawara M.

144th Audio Engineering Society Convention 2018 2018年01月

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

© 2018 Audio Engineering Society. All Rights Reserved. Music appreciation can be achieved with a variety of manners such as smartphones, portable music players, car audio, and high-end audio systems. Power consumption is one of the important issues for portable electronic devices and electric vehicles. In this paper, a power saving audio playback algorithm is proposed in restraint of perceptual distortion. An original music source is passed through filterbanks and is reconstructed after increasing or decreasing the narrow-band component in each channel. The channel-dependent manipulation is carefully done not to cause perceptual distortion. The feasibility of the proposed method is evaluated by both measuring consumption current while music playback and carrying out a listening test.

Scopus

その他リンク： https://www.scopus.com/inward/record.uri?partnerID=HzOxMe3b&scp=85050035197&origin=inward

Relationship among speech intelligibility and non-parametric distortion measures under various noise conditions 査読有り

M. Mizumachi, Y. Moriyama

Proc. Inter-noise Paper ID: 1324 2017年08月

Non-linear broadband beamformer with deep neural network 査読有り

M. Mizumachi, M. Origuchi, and Y. Nishijima

Proc. 24th International Congress on Sound and Vibration (ICSV24) Paper ID: 873 2017年07月

音響・超音波研究がひらいた今日の生活と今後の展望

水町光徳

電子情報通信学会誌 ( 電子情報通信学会 ) 100 ( 6 ) 419 - 423 2017年06月

加齢による聴力低下と発話変化の相互作用招待有り査読有り

水町光徳

日本音響学会誌 73 ( 5 ) 297 - 302 2017年05月

Discussion on Subjective Characteristics of High Resolution Audio 査読有り

M. Mizumachi, R. Yamamoto, K. Niyada

Proc. AES 142nd Convention EB02-07 2017年05月

Relationship among speech intelligibility and non - Parametric distortion measures under various noise conditions 査読有り

Mizumachi M., Moriyama Y.

INTER-NOISE 2017 - 46th International Congress and Exposition on Noise Control Engineering: Taming Noise and Moving Quiet 2017-January 2017年01月

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

© 2017 Institute of Noise Control Engineering. All rights reserved. Speech intelligibility under noisy environments is one of important issues in noise control engineering. It is necessary to subjectively evaluate speech intelligibility, although listening tests require enormous time and costs. Then, objective speech distortion measures have been investigated with parametric models under noisy environments. It is, however, difficult to design an all-purpose parametric distortion measure for severely distorted speech. This study aims at investigating the relationship among subjective speech intelligibility and non-parametric speech distortion measures. The non-parametric distortion measures include signal-to-noise ratio, spectral distortion, and cepstral distortion, which are calculated in each short-term frame. The spectral distortion is prepared in each octave frequency band. The short-term distortion measures yield the means and standard deviations of speech distortion. Multiple linear regression analyses are carried out, where the speech intelligibility and the objective distortion parameters are dependent and independent variables, respectively. The signal-to-noise ratio dominated speech intelligibility as expected. It is also confirmed that speech intelligibility was strongly influenced by the temporal variation of the cepstral distortion under non-stationary noise conditions.

Scopus

その他リンク： https://www.scopus.com/inward/record.uri?partnerID=HzOxMe3b&scp=85042060858&origin=inward

Non-linear broadband beamformer with deep neural network 査読有り

Mizumachi M., Origuchi M., Nishijima Y.

24th International Congress on Sound and Vibration, ICSV 2017 2017年01月

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

Speech enhancement is indispensable for human-to-machine speech interfaces, because speech signals are extremely fragile against acoustical interferences. Spatial filtering, that is, beamforming, is one of the popular solutions using multiple microphones. In the case of linear beamformers, a number of microphones are required to make the main lobe sharp at the desired direction. On the other hand, non-linear beamformers efficiently form the sharp main lobe compared with the linear beamformers. Kobatake et al. proposed a pioneering non-linear beamformer with a neural network. This method considered the narrowband optimization of the network structure, and succeeded in both achieving a sharper main lobe and decreasing grating lobes. Narrowband nonlinear beamformers have been steadily developed in the field of antenna and propagation. In this paper, a non-linear broadband beamformer with a deep neural network is proposed for speech enhancement. The proposed beamformer substitute a deep neural network for a conventional neural network, and non-equally-spaced microphone arrangement is employed for making grating lobes disappear. It is confirmed that the proposed beamformer can achieve ideal beam-pattern under a real environment.

Scopus

その他リンク： https://www.scopus.com/inward/record.uri?partnerID=HzOxMe3b&scp=85029461029&origin=inward

加齢による聴力低下と発話変化の相互作用招待有り査読有り

水町光徳

日本音響学会誌 ( 一般社団法人日本音響学会 ) 73 ( 5 ) 297 - 302 2017年01月

Advanced delay-and-sum beamformer with deep neural network 査読有り

M. Mizumachi and M. Origuchi

Proc. 22nd International Congress on Acoustics (ICA2016) Paper ID: 696 2016年09月

Speech intelligibility under in-car distant-talking environments 査読有り

Mizumachi M., Takuma S., Ohsugi I., Hamada Y., Nishi K.

Proceedings of the INTER-NOISE 2016 - 45th International Congress and Exposition on Noise Control Engineering: Towards a Quieter Future 389 - 393 2016年08月

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

© 2016, German Acoustical Society (DEGA). All rights reserved. Speech intelligibility has been widely investigated for various purposes. Hands-free speech communication without any close-talking microphone becomes popular inside a vehicle for the safety reason. It is necessary to carefully install microphones so that speech signals are clearly captured under noisy in-vehicle environments. In this study, the relationship between the intelligibility of distant-talking speech and acoustical interferences is investigated inside car cabins. Target speech signals are prepared as the utterance of Japanese 4-digits random sequences. The acoustical interferences consist of exterior and interior noises. Difficulty in speech perception is defined by the speech reception threshold, which corresponds to the sound pressure level of speech at the 50 % correct score in an articulation test. It is suggested that the interior blower noise caused by an air-conditioner is the most dominant interferences in speech intelligibility. For further investigation, multiple regression analyses have been carried out using subjective indexes related to speech intelligibility. It is found that speech intelligibility can be modeled with thickness of speech, loudness of speech, and loudness of noise.

Scopus

その他リンク： https://www.scopus.com/inward/record.uri?partnerID=HzOxMe3b&scp=84994560619&origin=inward

Speech intelligibility under in-car distant-talking environments 査読有り

M. Mizumachi, S. Takuma, I. Ohsugi, Y. Hamada, K. Nishi

Proc. Inter-noise Paper ID: 408 2016年08月

Perceptual restoration of intermittent speech using human speech-like noise 査読有り

M. Mizumachi, S. Imanaga, and T. Horiuchi

Proc. 23rd International Congress on Sound and Vibration (ICSV23) Paper ID: 1050 2016年07月

Subjective evaluation of high resolution audio through headphones 査読有り

M. Mizumachi, R. Yamamoto, and K. Niyada

Proc. AES 140th Convention Paper No. 9529 2016年06月

Singing support system using LPC residual wave for laryngectomee 査読有り

Himeno T., Tanaka Y., Mizumachi M., Matsui K.

2016 IEEE International Conference on Consumer Electronics, ICCE 2016 395 - 396 2016年03月

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）

© 2016 IEEE. In this paper, we propose a singing system to transmit the pseudo-vocal sound source to the vocal tract. The purpose is the quality improvement of singing voice when using an Electric-Larynx (EL), which is an alternative vocal apparatus of laryngectomee. Thus, we focus on the LPC residual wave, but it has a problem that the signal-power of the vowels /i/ and /u/ are largely reduced. Therefore we tried to improve the quality of sound by adding a signal-power compensation. We carried out listening experiments for the four sound sources «LPC residual wave», «LPC residual wave (signal-power Compensation)», «Rosenberg wave» and «EL». As a result, power-compensated LPC residual wave had the highest naturalness. However, the spectrum of the vowels /i/ differed greatly from the singing voice of the normal utterance.

DOI： 10.1109/ICCE.2016.7430662

Scopus

その他リンク： https://www.scopus.com/inward/record.uri?partnerID=HzOxMe3b&scp=84965132816&origin=inward

Perceptual restoration of intermittent speech using human speech-like noise 査読有り

Mizumachi M., Imanaga S., Horiuchi T.

ICSV 2016 - 23rd International Congress on Sound and Vibration: From Ancient to Modern Acoustics 2016年01月

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

Mobile phones have caused an explosive increase in packet distribution. It causes a serious problem, that is, packet loss. Packet loss concealment is indispensable for achieving smooth speech communication. As a packet loss concealment method on a client side, a waveform substitution is popular and standardized by ITU. The ITU-T G. 711 conceals packet loss by inserting a phase-adjusted amplitude-attenuating previous packet into each break, but cannot deal with long-term breaks over 60 ms, that is, burst loss. The authors have previously proposed an alternative packet loss concealment relying on a human auditory capability. This method does not aim at restoring a waveform of the intermittent speech signal, but achieves perceptual restoration relying on phonetic restoration. When a gap of a intermittent speech signal is filled up with a loud arbitrary signal, we can smoothly listen to restored speech even if some segments of the original speech signal are completely lost. There is a trade-off between smoothness of the restored speech and noisiness of the gap-filling signal. Previously, the gap-filling signal was composed of a harmonic complex and ambient noises. In this paper, a human speech-like noise is substituted for the gap-filling signal. The speech-like noise can be prepared by repeatedly overlapping short-term human speech signals. It is confirmed that the proposed gap-filling signal succeeds in reducing its noisiness.

Scopus

その他リンク： https://www.scopus.com/inward/record.uri?partnerID=HzOxMe3b&scp=84987899770&origin=inward

Subjective evaluation of high resolution audio through headphones 査読有り

Mizumachi M., Yamamoto R., Niyada K.

140th Audio Engineering Society International Convention 2016, AES 2016 2016年01月

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

Recently, high resolution audio (HRA) can be played back through portable devices, and spreads across musical genres and generation. It means that most people listen to HRA through headphones and earphones. In this study, perceptual discrimination among audio formats including HRA has been investigated using headphones. 36 subjects, who have a variety of audio and musical experiences in the wide age range from twenties to seventies, participated in listening tests. Headphones presentation is superior in discriminating the details to the loudspeakers presentation. It is, however, found that the headphones presentation is weak in reproducing presence and reality. Audio enthusiasts and musicians could significantly discriminate audio formats than ordinary listeners in both headphones and loudspeakers listening conditions.

Scopus

その他リンク： https://www.scopus.com/inward/record.uri?partnerID=HzOxMe3b&scp=84983565516&origin=inward

　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