• 전자공학회논문지 IE
• /
• v.49 no.2
• /
• pp.9-17
• /
• 2012

One of the problems on the application of the speech recognition system in the real world is the degradation of the performance by acoustical distortions. The most important source of acoustical distortion is the additive noise. This paper describes a spectral compensation technique based on a spectral peak enhancement scheme followed by an efficient noise subtraction scheme for noise robust speech recognition. The proposed methods emphasize the formant structure and compensate the spectral tilt of the speech spectrum while maintaining broad-bandwidth spectral components. The recognition experiments was conducted using noisy speech corrupted by white Gaussian noise, car noise, babble noise or subway noise. The new technique reduced the average error rate slightly under high SNR(Signal to Noise Ratio) environment, and significantly reduced the average error rate by 1/2 under low SNR(10 dB) environment when compared with the case of without spectral compensations.

• Journal of the Institute of Electronics Engineers of Korea CI
• /
• v.40 no.5
• /
• pp.289-297
• /
• 2003

The performance of a bimodal system is affected by the accuracy of the endpoint detection from the input signal as well as the performance of the speech recognition or lipreading system. In this paper, we propose the endpoint detection method which detects the endpoints from the audio and video signal respectively and utilizes the signal to-noise ratio (SNR) estimated from the input audio signal to select the reliable endpoints to the acoustic noise. In other words, the endpoints are detected from the audio signal under the high SNR and from the video signal under the low SNR. Experimental results show that the bimodal system using the proposed endpoint detector achieves satisfactory recognition rates, especially when the acoustic environment is quite noisy.

• The Journal of the Acoustical Society of Korea
• /
• v.8 no.4
• /
• pp.44-52
• /
• 1989

The UCERSS (Unit Circle Eigendecomposition Rational Signal Subspace) algorithm has extended MUSIC (MUltiple Signal Classification ) by using eigendecomposition on the unit circle in order to estimate incident angles of multiple wide band signals. The purpose of this thesis is to further extend the UCERSS to be able to estimate the direction of arrivals of multiple wide band signals in very low SNR . The wide band ESPRIT (Estimation of Signal Parameter via Rotational Invariance Technique) uses covariance difference matrices to reduce noise components. In this paper the wide band ESPRIT which combines the ideas of UCERSS and ESPRIT Is proposed. Computer simulation results Indicate that the performances of the proposed approaches are superior to those of the UCERSS in very low SNR.

• The Journal of the Acoustical Society of Korea
• /
• v.13 no.2
• /
• pp.50-59
• /
• 1994

This paper presents a speech recognition method utilizing third-order cumulants as a feature vector and a neural network for recognition. The use of higher-order cumulants provides desirable uncoupling between the gaussian noise and speech, which enables us to estimate the coefficients of AR model without bias. Unlike the conventional method using second-order statistics, the proposed one exhibits low bias even in SNR as low as 0 dB at the expense of higher variance. It is confirmed through computer simulation that recognition rate of korean single-vowels with the cumulant-based method is much higher than the results with the conventional method even in low SNR.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.36 no.12C
• /
• pp.764-773
• /
• 2011

Despite enormous performance gain with multi-antenna transmission in the single cell environment, its gain diminishes out in the multi-cell environment due to interference. It is also very hard to solve the efficient downlink beamforming with low complexity in multi-cell environment. First, this paper shows that the asymptotically sum rate optimal downlink beamformings at low and high SNR are maximum ratio transmit (MRT) and zero forcing (ZF) beamforming in the multi-cell system, respectively. Secondly, exploiting the asymptotically optimal downlink beamforming, we develop simple two types of near optimal downlink beamforming schemes having the form of minimum mean squared error (MMSE) beamforming obtained from the dual uplink problem. For each type, three different subclasses are also considered depending on the computational complexity. The simulation results show that the proposed near optimum algorithms provide the trade-off between the complexity and the performance.

• Current Optics and Photonics
• /
• v.4 no.1
• /
• pp.69-80
• /
• 2020

A Rayleigh Lidar used for wind detection works by transmitting laser pulses to the atmosphere and receiving backscattering signals from molecules. Because of the weak backscattering signals, a lidar usually uses a high sensitivity photomultiplier as detector and photon counting technology for signal collection. The capturing of returned extremely weak backscattering signals requires the lidar to work on dark background with a long time accumulation to get high signal-to-noise ratio (SNR). Because of the strong solar background during the day, the SNR of lidar during daytime is much lower than that during nighttime, the altitude and accuracy of detection are also restricted greatly. Therefore this article describes an ultra-narrow bandwidth filter (UNBF) that has been developed on 354.7 nm wavelength of laser. The UNBF is used for suppressing the strong solar background that degrades the performance of Rayleigh wind lidar during daytime. The optical structure of UNBF consists of an interference filter (IF), a low resolution Fabry-Perot interferometer (FPI) and a high resolution FPI. The parameters of each optical component of the UNBF are presented in this article. The transmission curve of the aligned UNBF is measured with a tunable laser. Contrasting the result of with-UNBF and with-IF shows that the solar background received by a Licel transient recorder decreases by 50~100 times and that the SNR with-UNBF was improved by 3 times in the altitude range (35 km to 40 km) compared to with-IF at 10:26 to 10:38 on August 29, 2018. By the SNR comparison at four different times of one day, the ratio-values are larger than 1 over the altitude range (25~50 km) in general, the results illustrate that the SNR with-UNBF is better than that with-IF for Rayleigh Lidar during daytime and they demonstrate the effective improvements of solar background restriction of UNBF.

• Journal of Communications and Networks
• /
• v.2 no.1
• /
• pp.58-68
• /
• 2000

In this paper we introduce the correlation-aided distributed sample acquisition (CDSA) scheme for fast cell search in IMT-2000 W-CDMA cellular system. The proposed scheme incorporates the state symbol correlation process into the comparison-correction based synchronization process of the original DSA scheme to enable fast acquisition even under very poor channel environment. for its realization, each mobile station (MS) has to store in its memory a set of state sample sequences. which are determined by the long-period scrambling sequences used in the system and the sampling interval of the state samples. CDSA based cell search is carried out in two stages : First, the MS first acquires the slot timing by using the primary synch code (PSC) and then identifies the igniter code which conveys the state samples of the current cell . Secondly. the MS identifies the scrambling code and frame timing by taking the comparison-correction based synchronization approach and, if the identification is not done satisfactorily within preset time. it initiates the state symbol correlation process which correlates the received symbol sequence with the pre-stored state sample sequences for a successful identification. As the state symbol SNR is relatively high. the state symbol correlation process enables reliable synchronization even in very low chip-SNR environment. Simulation results show that the proposed CDSA scheme outperforms the 3GPP 3-step approach, requiring the signal power of about 7 dB less for achieving the same acquisition time performance in low-SNR environments. Furthermore, it turns out very robust in the typical synchronization environment where large frequency offset exists.

• Journal of the Korean Society of Marine Environment & Safety
• /
• v.15 no.3
• /
• pp.243-250
• /
• 2009

The expression of signal to noise ratio (SNR) is derived for the individual echo sounder and for the general specifications of commercial echo sounder. The SNR is depicted by several factors on transmitting and receiving, sound propagation, scattering by fishes, and mainly self ship noise. The detection ranges of echo sounders in depth and breadth are derived by finding the border of an acceptable SNR. The detection ranges are computed for the echo sounders of individual and general specifications. Generally, the detection range is lager for low frequencies. Increasing an electrical power and a diameter of transducer is not as effective for high frequencies as for low frequencies to increase the detection range. The theoretical results of the detection range can be applied to design low electrical power echo sounder for coastal use, to select an appropriate echo sounder, to know the capability of a sounder, and to interpret echograms. A method for evaluating and designing low electrical power echo sounder is developed. A universal diagram for the basic design of the low electrical power echo sounder where SNR is plotted against frequency is developed. An appropriate frequency, transducer diameter, and beam width are easily determined by using the diagram.

• Journal of Integrative Natural Science
• /
• v.4 no.3
• /
• pp.238-242
• /
• 2011

For any pattern matching based algorithm in WLAN environment, the characteristics of signal to noise ratio(SNR) to multiple access points(APs) are utilized to establish database in the training phase, and in the estimation phase, the actual two dimensional coordinates of mobile unit(MU) are estimated based on the comparison between the new recorded SNR and fingerprints stored in database. The system that uses the artificial neural network(ANN) falls in a local minima when it learns many nonlinear data, and its classification accuracy ratio becomes low. To make up for this risk, the SVM/ANN hybrid algorithm is proposed in this paper. The proposed algorithm is the method that ANN learns selectively after clustering the SNR data by SVM, then more improved performance estimation can be obtained than using ANN only and The proposed algorithm can make the higher classification accuracy by decreasing the nonlinearity of the massive data during the training procedure. Experimental results indicate that the proposed SVM/ANN hybrid algorithm generally outperforms ANN algorithm.

• ETRI Journal
• /
• v.31 no.4
• /
• pp.451-453
• /
• 2009

This letter proposes the use of vowel sound detection for voice activity detection. Vowels have distinctive spectral peaks. These are likely to remain higher than their surroundings even after severe corruption. Therefore, by developing a method of detecting the spectral peaks of vowel sounds in corrupted signals, voice activity can be detected as well even in low signal-to-noise ratio (SNR) conditions. Experimental results indicate that the proposed algorithm performs reliably under various noise and low SNR conditions. This method is suitable for mobile environments where the characteristics of noise may not be known in advance.