• The Transactions of The Korean Institute of Electrical Engineers
• /
• v.61 no.12
• /
• pp.1920-1925
• /
• 2012

In this paper, we propose a novel method to improve the performance of the improved minima controlled recursive averaging (IMCRA) in non-stationary noisy environment. The conventional IMCRA algorithm efficiently estimate the noise power by averaging past spectral power values based on a smoothing parameter that is adjusted by the signal presence probability in frequency subbands. Since the minimum of smoothing parameter is defined as 0.85, it is difficult to obtain the robust estimates of the noise power in non-stationary noisy environments that is rapidly changed the spectral characteristics such as babble noise. For this reason, we proposed the modified IMCRA, which adaptively estimate and updata the noise power according to the noise type classified by the Gaussian mixture model (GMM). The performances of the proposed method are evaluated by perceptual evaluation of speech quality (PESQ) and composite measure under various environments and better results compared with the conventional method are obtained.

• Phonetics and Speech Sciences
• /
• v.2 no.2
• /
• pp.117-121
• /
• 2010

The This study aimed to evaluate the effects of the voice therapy we operated to the patients with age-related dysphonia. Thirty four participants who were diagnosed as age-related dysphonia in laryngoscopic finding from January, 2009 to December, 2009 completed the study. The participants were aged from 60 to 82 years old with a mean age of 70.6. All participants had received the abdominal breath technique, SKHPIP with laughter, and basic vocal training with description of their problem, the length of which ranged from four sessions to twelve sessions. We executed the videostroboscopy to compare the aspect of voicing change and the perceptual assessment, voice range profile, acoustic and aerodynamic measures to identify change of voice. Participants had glottal gap due to incomplete glottic closure during voicing on the pretest. After they took the voice therapy, the glottic gap became narrow and rough and breathy voice was reduced. There were significant difference in acoustic and aerodynamic measures. Jitter, Shimmer, MFR were reduced and MPT, Psub were increased(p<.05). Participants' pitch range and intensity range were increased on the posttest performance after taking voice therapy. Especially, most of them were showed that pitch range was increased significantly in high frequency area. The results of this investigation indicate that the voice therapy using abdominal breath, SKHPIP, and exercise together is effective for the patients who have age-related dysphonia to improve their voice quality. We recommend to apply this technique to functional voice disorders who are showed glottal gap.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2015.07a
• /
• pp.418-420
• /
• 2015

본 논문은 베이지안 비음수 행렬 인수분해 (Bayesian nonnegative matrix factorization, BNMF) 기반의 음성 강화 기법에서 음성과 잡음 성분의 latent source 수에 따른 강화성능에 대해 서술한다. BNMF 기반의 음성 강화 기법은 입력 신호를 서브 신호들의 합으로 분해한 후, 잡음 성분을 제거하는 방식으로 그 성능이 기존의 NMF 기반의 방법들보다 우수한 것으로 알려져 있다. 그러나 많은 계산량과 latent source 의 수에 따라 성능의 차이가 있다는 단점이 있다. 이러한 단점을 개선하기 위해 본 논문에서는 BNMF 기반의 음성 강화 기법에서 최적의 latent source 개수를 찾기 위한 실험을 진행하였다. 실험은 잡음의 종류, 음성의 종류, 음성과 잡음의 latent source 의 개수, 그리고 SNR 을 바꿔가며 진행하였고, 성능 평가 방법으로 PESQ (perceptual evaluation of speech quality) 를 이용하였다. 실험 결과, 음성의 latent source 개수는 성능에 영향을 주지 않지만, 잡음의 latent source 개수는 많을수록 성능이 좋은 것으로 확인되었다.

• The Transactions of The Korean Institute of Electrical Engineers
• /
• v.64 no.12
• /
• pp.1756-1760
• /
• 2015

This paper focuses on a priori signal to noise ratio (SNR) estimation method for the speech enhancement. There are many researches for speech enhancement with several ambient noise cancellation methods. The method based on spectral subtraction (SS) which is widely used in noise reduction has a trade-off between the performance and the distortion of the signals. So the need of adaptive method like an estimated a priori SNR being able to making a high performance and low distortion is increasing. The decision directed (DD) approach is used to determine a priori SNR in noisy speech signals. A priori SNR is estimated by using only the magnitude components and consequently follows a posteriori SNR with one frame delay. We propose a modified a priori SNR estimator and the weighted rational transfer function for speech enhancement with wind noises. The experimental result shows the performance of our proposed estimator is better Perceptual Evaluation of Speech Quality scores (PESQ, ITU-T P.862) compare to the conventional DD approach-based systems and different noise reduction methods.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.17 no.1
• /
• pp.32-40
• /
• 2013

In this paper, we propose a multi-channel speech enhancement algorithm using beamforming and direction-of-arrival (DOA)-based masking. The proposed algorithm enhances noisy speech basically by the linearly constrained minimum variance (LCMV) algorithm and then a mel-scale Wiener filter designed using DOA-based masking is applied to remove still remaining noises. To improve the performance, we optimize the learning rate of the adaptive filters in LCMV and the DOA threshold to detect target speech spectrum. As performance indices, the perceptual evaluation of speech quality (PESQ) score and output SNRs are measured. Experimantal results show that the proposed algorithm outperforms the conventional LCMV beamformer by 0.09 in PESQ score and 5.75 dB in output SNR, respectively.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2018.05a
• /
• pp.335-337
• /
• 2018

음성향상기법은 음성에 포함된 잡음이나 잔향을 제거하는 기술로써 마이크로폰으로 입력된 음성신호는 잡음이나 잔향에 의해 왜곡되어지므로 음성인식, 음성통신 등의 음성신호처리 기술의 핵심 기술이다. 이전에는 음성신호와 잡음신호 사이의 통계적 정보를 이용하는 통계모델 기반의 음성향상기법이 주로 사용되었으나 통계 모델 기반의 음성향상기술은 정상 잡음 환경과는 달리 비정상 잡음 환경에서 성능이 크게 저하되는 문제점을 가지고 있었다. 최근 머신러닝 기법인 심화신경망 (DNN, deep neural network)이 도입되어 음성 향상 기법에서 우수한 성능을 내고 있다. 심화신경망을 이용한 음성 향상 기법은 다수의 은닉 층과 은닉 노드들을 통하여 잡음이 존재하는 음성 신호와 잡음이 존재하지 않는 깨끗한 음성 신호 사이의 비선형적인 관계를 잘 모델링하였다. 이러한 심화신경망 기반의 음성향상기법을 향상 시킬 수 있는 방법 중 하나인 강화학습을 적용하여 기존 심화신경망 대비 성능을 향상시켰다. 강화학습이란 대표적으로 구글의 알파고에 적용된 기술로써 특정 state에서 최고의 reward를 받기 위해 어떠한 policy를 통한 action을 취해서 다음 state로 나아갈지를 매우 많은 경우에 대해 학습을 통해 최적의 action을 선택할 수 있도록 학습하는 방법을 말한다. 본 논문에서는 composite measure를 기반으로 reward를 설계하여 기존 PESQ (Perceptual Evaluation of Speech Quality) 기반의 reward를 설계한 기술 대비 음성인식 성능을 높였다.

• The Journal of Korea Robotics Society
• /
• v.16 no.2
• /
• pp.94-102
• /
• 2021

This study addresses a method for 3D reconstruction using acoustic data with heterogeneous sonar devices: Forward-Looking Multibeam Sonar (FLMS) and Profiling Sonar (PS). The challenges in sonar image processing are perceptual ambiguity, the loss of elevation information, and low signal to noise ratio, which are caused by the ranging and intensity-based image generation mechanism of sonars. The conventional approaches utilize additional constraints such as Lambertian reflection and redundant data at various positions, but they are vulnerable to environmental conditions. Our approach is to use two sonars that have a complementary data type. Typically, the sonars provide reliable information in the horizontal but, the loss of elevation information degrades the quality of data in the vertical. To overcome the characteristic of sonar devices, we adopt the crossed installation in such a way that the PS is laid down on its side and mounted on the top of FLMS. From the installation, FLMS scans horizontal information and PS obtains a vertical profile of the front area of AUV. For the fusion of the two sonar data, we propose the probabilistic approach. A likelihood map using geometric constraints between two sonar devices is built and a monte-carlo experiment using a derived model is conducted to extract 3D points. To verify the proposed method, we conducted a simulation and field test. As a result, a consistent seabed map was obtained. This method can be utilized for 3D seabed mapping with an AUV.

• Journal of Information Processing Systems
• /
• v.18 no.2
• /
• pp.258-267
• /
• 2022

Weak over-the-horizon perception and blind spot are the main problems in intelligent connected vehicles (ICVs). In this paper, a V2V image transmission-based road condition warning method is proposed to solve them. The encoded road emergency images which are collected by the ICV are transmitted to the on-board unit (OBU) through Ethernet. The OBU broadcasts the fragmented image information including location and clock of the vehicle to other OBUs. To satisfy the channel quality of the V2X communication in different times, the optimal fragment length is selected by the OBU to process the image information. Then, according to the position and clock information of the remote vehicles, OBU of the receiver selects valid messages to decode the image information which will help the receiver to extend the perceptual field. The experimental results show that our method has an average packet loss rate of 0.5%. The transmission delay is about 51.59 ms in low-speed driving scenarios, which can provide drivers with timely and reliable warnings of the road conditions.

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.9 no.5
• /
• pp.1407-1413
• /
• 2008

Individual consumer characteristics change as the individual lifestyle changes. Each individual's unique lifestyle influences substantially the individual consumer behavior. Individual consumer's behavioral pattern varies significantly depending on the individual consumer's use of credit cards. Each individual's lifestyle or individual's perception of credit cards' brands and perceptual difference in qualities of credit cards' brands influence the individual's credit card selection. Credit card companies have been doing several researches in analyzing credit card users' lifestyle characteristics and consumer behavioral characteristics. However, researches on the relation of model structure among variables such as individual lifestyle, credit cards brand, quality perception of credit cards and credit card selection are not quite noticeable. Therefore, in this research we aim at providing a theoretical foundation with credit card companies by analyzing the relation of model structure among these factors.

• The Journal of the Acoustical Society of Korea
• /
• v.19 no.7
• /
• pp.42-47
• /
• 2000

This paper describes a real-time implementation of a GSM-EFR (Global System for Mobil communications Enhanced Full Rate) speech coder using OakDSP core; a 16bit fixed-point Digital Signal Processor (DSP) by DSP Group, Inc. The real-time implemented speech coder required about 24MIPS for computation and 7.06K words and 12.19K words for code and data memory, respectively. The implemented GSM-EFR speech coder passes all of test vectors provided by ETSI (European Telecommunication Standard Institute), and perceptual speech quality measurement using MNB algorithm shows that the quality of the GSM-EFR speech coder is similar to the one of 32kbps ADPCM. The real-time implemented GSM-EFR speech coder which is the highest bit-rate mode of the GSM-AMR speech coder will be used as the basic structure of the GSM-AMR speech coder which is embedded in MODEM ASIC of IMT2000 asynchronous mode mobile station.