• The Journal of the Acoustical Society of Korea
• /
• 제23권1E호
• /
• pp.16-18
• /
• 2004

There have been numerous studies on the enhancement of the noisy speech signal. In this paper, I propose a completely new speech enhancement method, that is, a filtering of a dissonant frequency combined with noise reduction algorithm. The simulation results indicate that the proposed method provides a significant gain in audible improvement compared with the conventional method. Therefore if the proposed enhancement scheme is used as a pre-filter, the perceptual quality of speech is greatly enhanced.

• ETRI Journal
• /
• 제38권6호
• /
• pp.1190-1196
• /
• 2016

We propose a new bandpass filter (BPF)-based online channel normalization method to dynamically suppress channel distortion when the speech and channel noise components are unknown. In this method, an adaptive modulation frequency filter is used to perform channel normalization, whereas conventional modulation filtering methods apply the same filter form to each utterance. In this paper, we only normalize the two mel frequency cepstral coefficients (C0 and C1) with large dynamic ranges; the computational complexity is thus decreased, and channel normalization accuracy is improved. Additionally, to update the filter weights dynamically, we normalize the learning rates using the dimensional power of each frame. Our speech recognition experiments using the proposed BPF-based blind channel normalization method show that this approach effectively removes channel distortion and results in only a minor decline in accuracy when online channel normalization processing is used instead of batch processing

• 지질공학
• /
• 제27권4호
• /
• pp.513-518
• /
• 2017

Groundroll is a coherent noise showing dispersive behavior in land seismic-reflection records and its rejection has been a stubborn problem in data processing because they severely degrade the continuities and resolution of reflection signals. Conventional processing schemes of attenuating noises are the kind of frequency filtering (i.e., bandpass and f-k) that uses the Fourier transform (FT) along the entire trace in the time domain. To suppress them in this study, 1D wavelet filtering (WT) that can control time-varying frequency properties is tested and performed in the land-based synthetic and field seismic data. The results are compared to the ones from conventional filtering techniques in terms of continuities and resolution of reflection events. This filtering technique enhanced the reflection events by effectively eliminating the dispersive groundroll and random noises with control of time-scale function on wavelet domain.

• 한국콘텐츠학회:학술대회논문집
• /
• 한국콘텐츠학회 2009년도 춘계 종합학술대회 논문집
• /
• pp.648-651
• /
• 2009

다양한 통신시스템 설계에 사용되는 다중 스텁 도파관 구조가 주파수 필터링 특성을 가진 것을 모드정합법을 이용해 보인다. 주파수 필터링 특성을 가진 다중 스텁 도파관 구조의 통과 대역과 차단 대역을 자동으로 설계하기 위해 적합도 함수를 정의하고 유전 알고리즘으로 구현 가능함을 보인다. 스텁 도파관 구조가 주파수 필터링 특성을 가진 것을 수치해석적으로 보이고 다양한 매개변수에 대한 특성변화를 관찰한다.

• JSTS:Journal of Semiconductor Technology and Science
• /
• 제13권2호
• /
• pp.170-183
• /
• 2013

This paper gives an overview of fractional-N phase-locked loops (PLLs) with practical design perspectives focusing on a ${\Delta}{\Sigma}$ modulation technique and a finite-impulse response (FIR) filtering method. Spur generation and nonlinearity issues in the ${\Delta}{\Sigma}$ fractional-N PLLs are discussed with simulation and hardware results. High-order ${\Delta}{\Sigma}$ modulation with FIR-embedded filtering is considered for low noise frequency generation. Also, various architectures of finite-modulo fractional-N PLLs are reviewed for alternative low cost design, and the FIR filtering technique is shown to be useful for spur reduction in the finite-modulo fractional-N PLL design.

• Structural Engineering and Mechanics
• /
• 제36권6호
• /
• pp.697-714
• /
• 2010

This paper presents an improved version of the Hilbert-Huang transform (HHT) for the modal evaluation of structural systems or signals. In this improved HHT, a well-designed bandpass filter is used as preprocessing to separate and determine each mode of the signal for solving the inherent modemixing problem in HHT (i.e., empirical mode decomposition, EMD, associated with the Hilbert transform). A screening process is then applied to remove undesired intrinsic mode functions (IMFs) derived from the EMD of the signal's mode. A "best" IMF is selected in each screening process that utilizes the orthogonalization coefficient between the signal's mode and its IMFs. Through mode-by-mode signal filtering, parameters such as the modal frequency can be evaluated accurately when compared to the theoretical value. Time history of the identified modal frequency is available. Numerical results prove the efficiency of the proposed approach, showing relative errors 1.40%, 2.06%, and 1.46%, respectively, for the test cases of a benchmark structure in the lab, a simulated time-varying structural system, and of a linear superimposed cosine waves.

• 한국해양공학회지
• /
• 제24권1호
• /
• pp.166-171
• /
• 2010

In this paper, the front-end of a digital receiver with a 25 kHz carrier frequency, 5 kHz symbol rate, and any excess-bandwidth is designed using two basic facts. The first is known as the uniform sampling theorem, which states that the sampled sequence might not suffer from aliasing even if its sampling rate is lower than the Nyquist sampling rate if the analog signal is a bandpass one. The other fact is that if the sampling rate is 4 times the center frequency of the sampled sequence, the front-end processing complexity can be dramatically reduced due to the half of the sampled sequence to be multiplied by zero in the demixing process. Furthermore, the designed front-end is simplified by introducing sub-filters and sub-sampling sequences. The designed front-end is composed of an A/D converter, which takes samples of a bandpass filtered signal at a 20 kHz rate; a serial-to-parallel converter, which converts a sampled bandpass sequence to 4 parallel sub-sample sequences; 4 sub-filter blocks, which act as a frequency shifter and lowpass filter for a complex sequence; 4 synchronized switches; and 2 adders. The designed front-end dramatically reduces the computational complexity by more than 50% for frequency shifting and lowpass filtering operations since a conventional front-end requires a frequency shifting and two lowpass filtering operations to get one lowpass complex sample, while the proposed front-end requires only four filtering operation to get four lowpass complex samples, which is equivalent to one filtering operation for one sample.

• 한국통신학회논문지
• /
• 제17권9호
• /
• pp.1029-1039
• /
• 1992

본 논문은 한글인식 방법중 주파수 변환영역에서 행하는 필터링의 방법을 제시한 것이다. 한글 문자패턴을 2차원 변조처리하여 공간위치에 의존적인 한글자소의 위상문자적 특징을 주파수 영역으로 사상하였고, 이때 변조 주파수를 정규화함으로서 주파수 영역에서 문자의 크기를 정규화 한다. 또한 한글의 각 자소를 발생위치에 따라 분류하여 표준패턴으로 설정하고, 설정된 각 자소의 표준패턴을 자소필터로 사용하여 주파수 영역으로 사상된 문자패턴을 필터링하였다. 한글자소의 분별 파라미터로는 정규화된 상호상관함수와 필터링 결과로 부터 유도된 코히어런스 함수를 산출하여 분별기준을 설정하였으며, 그 결과로서 문자크기의 변화, 자소의 융착, 제한적인 자획의 유실이나 잡음의 혼입등의 요인이 발생하였음 경우에도 한글자소의 분별이 가능하였고 또한 표준패턴을 설정함에 따라 숫자, 영문자등의 분별에도 적용할 수 있었다.

• 대한전기학회:학술대회논문집
• /
• 대한전기학회 1997년도 추계학술대회 논문집 학회본부
• /
• pp.603-605
• /
• 1997

A new algorithm using 1-dimensional wavelet transform for autofocusing of optical instrument has been developed. Previous studies based on the conventional frequency analysis have shown that as the lens-object distance approaches the optimum value, the high frequency energy in the corresponding image shows a consistent increase. However, as conventional frequency analysis techniques hide spatial distribution of each band energy, shape information in the original signal cannot be easily utilized. In this paper, a newly devised wavelet based focus measuring scheme is presented. Unlike other frequency domain analysis techniques that simply produce "frequency-only" spectra, wavelet analysis provides a "time-frequency" localized view of a given signal. As a result, both frequency band filtering and spatial distribution filtering can easily be realized. Depending on the proposed focus quality measuring algorithm, a fast and reliable automatic focus adjustment of optical devices could be implemented.

• 대한설비관리학회지
• /
• 제23권4호
• /
• pp.57-64
• /
• 2018

User-based collaborative filtering is a method of recommending an item to a user based on the preference of the neighbor users who have similar purchasing history to the target user. User-based collaborative filtering is based on the fact that users are strongly influenced by the opinions of other users with similar interests. Item-based collaborative filtering is a method of recommending an item by comparing the similarity of the user's previously preferred items. In this study, we create a recommendation model using user-based collaborative filtering and item-based collaborative filtering with consumer's consumption data. Collaborative filtering is performed by using RFM (recency, frequency, and monetary) technique with purchasing data to recommend items with high purchase potential. We compared the performance of the recommendation system with the purchase amount and the performance when applying the RFM method. The performance of recommendation system using RFM technique is better.