• The Journal of the Acoustical Society of Korea
• /
• v.12 no.1E
• /
• pp.32-38
• /
• 1993

본 논문에서는 저 전송속도의 음성 부호화기를 제안하였고 컴퓨터 시뮬레이션을 통하여 성능분석과 유연성을 입증하였다. 제안된 부호화 방식은 입력 음성신호의 Stationarity에 따라 전송 프레임의 길이를 가변하고, 전송 프레임의 대표적인 특징 벡터를 Vector Quatization으로 부호화하였다. 제안된 부호화 방식에서 특징 벡터열은 입력 음성신호를 샘플단위로 Prewindowed RLS Lattice 알고리즘을 통해 구한 PARCOR 계수로 구성된다. 입력 음성신호는 Subsegment로 분할되고, 각 Subsegment에서 대표적인 PARCOR 계수를 구한다. Likelihood Ratio Distortion Measure를 사용하여 유사도에 따라 Subsegment를 병합함으로써 전송프레임을 결정한다. 컴퓨터 시뮬레이션 결과로부터 제안된 VTEL 음성 부호화 방식은 좋은 음질을 유지하면서 전체 전송속도를 크게 줄일 수 있다.

• Journal of the Korean Institute of Telematics and Electronics
• /
• v.26 no.7
• /
• pp.146-156
• /
• 1989

This paper discusses how to choose speech parameters and decision logics to improve the performance of speaker recognition systems. It also considers the influence of the reference patterns on the speaker recognition. It is observed from the performance analysis based on LPSs, PARCOR coefficients and LPC-cepstrum coefficients that LPC-cepstrum coefficients are superior to the others in speaker recognition without regard to the reference patterns. In order to improve the recognition performance, a new decision logic is proposed based on a generalized-distance concept. It differs from the existing methods in that it considers the statistics of customer and impostors at the same time. It turns out from a speaker verification test that the proposed decision logic ferforms better than the existing ones.

• Journal of the Korea Society of Computer and Information
• /
• v.16 no.12
• /
• pp.187-196
• /
• 2011

Induction motors play a vital role in aeronautical and automotive industries so that many researchers have studied on developing a fault detection and classification system of an induction motor to minimize economical damage caused by its fault. With this reason, this paper extracts robust feature vectors from the normal/abnormal vibration signals of the induction motor in noise circumstance: partial autocorrelation (PARCOR) coefficient, log spectrum powers (LSP), cepstrum coefficients mean (CCM), and mel-frequency cepstrum coefficient (MFCC). Then, we classified different types of faults of the induction motor by using the extracted feature vectors as inputs of a neural network. To find optimal feature vectors, this paper evaluated classification performance with 2 to 20 different feature vectors. Experimental results showed that five to six features were good enough to give almost 100% classification accuracy except features by CCM. Furthermore, we considered that vibration signals could include noise components caused by surroundings. Thus, we added white Gaussian noise to original vibration signals, and then evaluated classification performance. The evaluation results yielded that LSP was the most robust in noise circumstance, then PARCOR and MFCC followed by LSP, respectively.

• Proceedings of the Acoustical Society of Korea Conference
• /
• 1993.06a
• /
• pp.121-124
• /
• 1993

This paper is a study on the word recognition using neural network. A limited vocabulary, speaker independent, isolated word recognition system has been built. This system recognizes isolated word without performing segmentation, phoneme identification, or dynamic time wrapping. It needs a static pattern approach to recognize a spatio-temporal pattern. The preprocessing only includes preceding and tailing silence removal, and word length determination. A LPC analysis is performed on each of 24 equally spaced frames. The PARCOR coefficients plus 3 other features from each frame is extracted. In order to simplify a structure of neural network, we composed binary code form to decrease output nodes.

• The Journal of the Acoustical Society of Korea
• /
• v.5 no.3
• /
• pp.21-27
• /
• 1986

LSP 방식은 음성의 주파수 특성을 포함하는 공진 주파수를 낮은 부분과 SHB은 부분의 주파수 로 표시되는 선스펙트럼쌍 계수를 구하는 방법이다. 본 논문은 LSP 방식을 사용하여 한국어의 기본 모 음 7개를 대상으로 하여 분석하고 LSP 계수를 구하는 Algorithm을 개발하였으며 PARCOR 방식과 비 교하였다. 실험 결과 LSP 방식의 연산량이 PARCO 방식의 연산량보다 약 1/2정도로 적음을 알 수 있었 고 Hardware 구성 시에 있어서도 경제적임을 알 수 있었다. 그리고 LSP는 계수 모음의 종류에 따라 각 기 다른 공진 주파수, 대역폭을 나타내기 때문에 음성 합성이나 음성 인식 분야에 있어 기초 자료로 이 용할 수 있을 것으로 사료된다.

• The Journal of the Acoustical Society of Korea
• /
• v.5 no.2
• /
• pp.19-28
• /
• 1986

DSP LSE을 이용하여 디지털 DTMF 수신기를 구현하는 방법으로는 IIR 디지털 필터, Counter 방법, DFT 방법, FFT 방법 및 PARCOR 방법등이 제안되어 왔다. 그 중에서도 IIR 디지털 필터를 이용 한 방법은 기존의 아나로그 DTME 수신기를 그대로 디지털화 한 것이기 때문에 성능이 제일 우수한 것 으로 알려져 있다. 그러나 IIR 디지털 필터를 이용하여 그것을 구현할 때 필터의 계수, roundoff 잡음, overflow 등 고려해야 할 사항이 많다. 본 논문에서는 이러한 문제점들을 해결하면서 CCITT 사양들을 만족하는 디지털 DTMF 수신기 구현에 관한 연구결과를 제시하였다. DSP LSI을 이용해서 수신기를 hardware 제작할 때 이 결과들을 수정없이 이용할 수 있다고 기대된다.

• The Journal of the Acoustical Society of Korea
• /
• v.12 no.4
• /
• pp.57-62
• /
• 1993

본 논문은 신경망을 이용한 우리말 모음에 대한 인식 연구이다. 음성을 나누거나. 음소별 인식이나, 시간 신축 방법을 사용하지 않고 모음을 인식하였다. 식나의 변화에 따른 음성의 변화를 정적인 음성으로 취급하였다. 10개로 균등히 나눈 프레임에 각 프레임마다 10차의 PARCOR계수를 추출하였다. 신경망의 구조를 간단히 하기 위해서 단모음과 복모음을 구분하여 학습시켰으며, 출력 노드의 수를 감소시키기 위해 이진 코드 형태로 구성하였다.

• The Journal of the Acoustical Society of Korea
• /
• v.13 no.6
• /
• pp.5-14
• /
• 1994

This paper is about an experiment of speaker-independent automatic Korean spoken words recognition using Multi-Layered Perceptron and Error Back-propagation algorithm. The object words are 50 citynames of D.D.D local numbers. 43 of those are 2 syllables and the rest 7 are 3 syllables. The words were not segmented into syllables or phonemes, and some feature components extracted from the words in equal gap were applied to the neural network. That led independent result on the speech duration, and the PARCOR coefficients calculated from the frames using linear predictive analysis were employed as feature components. This paper tried to find out the optimum conditions through 4 differerent experiments which are comparison between total and pre-classified training, dependency of recognition rate on the number of frames and PAROCR order, recognition change due to the number of neurons in the hidden layer, and the comparison of the output pattern composition method of output neurons. As a result, the recognition rate of $89.6\%$ is obtaimed through the research.

• Proceedings of the Acoustical Society of Korea Conference
• /
• 1994.06c
• /
• pp.91-94
• /
• 1994

This paper is a study on the design and implementation of the Korean Tet-to-Speech system. In this paper, parameter symthesis method is chosen for speech symthesis method and PARCOR coeffient, one of the LPC analysis, is used as acoustic parameter, We use a diphone as synthesis unit, it include a basic naturalness of human speech. Diphone DB is consisted of 1228 PCM files. LPC synthesis method has defect that decline clearness of synthesis speech, during synthesizing unvoiced sound In this paper, we improve clearness of synthesized speech, using residual signal as ecitation signal of unvoiced sound. Besides, to improve a naturalness, we control the prosody of synthesized speech through controlling the energy and pitch pattern. Synthesis system is implemented at PC/486 and use a 70Hz-4.5KHz band pass filter for speech imput/output, amplifier and TMS320c30 DSP board.

• Journal of the Korean Institute of Telematics and Electronics B
• /
• v.32B no.12
• /
• pp.1635-1643
• /
• 1995

This paper proposes an algorithm for machine recognition of phonemes in continuously spoken Korean. The proposed algorithm is a static strategy neural network. The algorithm uses, at the stage of training neurons, features such as the rate of zero crossing, short-term energy, and either PARCOR or auditory-like perceptual linear prediction(PLP) but not both, covering a time of 171ms long. Numerical results show that the algorithm with PLP achieves approximately the frame-based phoneme recognition rate of 99% for small vocabulary recognition experiments. Based on this it is concluded that the proposed algorithm with PLP analysis is effective in phoneme recognition.