• Proceedings of the KSPS conference
• /
• 2003.05a
• /
• pp.119-122
• /
• 2003

In general, triangular shape filters are used in the filter bank when we get the MFCCs from the spectrum of speech signal. In [1], a new feature extraction approach is proposed, which uses specific filter shapes in the filter bank that are obtained from the spectrum of training speech data. In this approach, principal component analysis technique is applied to the spectrum of the training data to get the filter coefficients. In this paper, we carry out speech recognition experiments, using the new approach given in [1], for a large amount of telephone speech data, that is, the telephone speech database of Korean connected digit released by SITEC. Experimental results are discussed with our findings.

• The Transactions of The Korean Institute of Electrical Engineers
• /
• v.67 no.6
• /
• pp.794-798
• /
• 2018

There is a growing need for a care system that can continuously monitor, manage and effectively relieve stress for modern people. In recent years, mobile healthcare devices capable of measuring heart rate have become popular, and many stress monitoring techniques using heart rate variability analysis have been actively proposed and commercialized. In addition, respiratory biofeedback methods are used to provide stress relieving services in environments using mobile healthcare devices. In this case, breathing information should be measured well to assess whether the user is doing well in biofeedback training. In this study, we extracted the heart beat interval signal from the PPG and used the oscillator based notch filter based on the IIR band pass filter to track the strongest frequency in the heart beat interval signal. The respiration signal was then estimated by filtering the heart beat interval signal with this frequency as the center frequency. Experimental results showed that the number of breathing could be measured accurately when the subject was guided to take a deep breath. Also, in the timeing measurement of inspiration and expiration, a time delay of about 1 second occurred. It is expected that this will provide a respiratory biofeedback service that can assess whether or not breathing exercise are performed well.

• Transactions of the Korean Society for Noise and Vibration Engineering
• /
• v.21 no.1
• /
• pp.31-40
• /
• 2011

This paper examines the dynamic characteristics of a human carpal muscle through theoretical analysis and experiment. The carpal muscle was modeled as a 1-DOF vibration system and vibration response due to a ramp function force was calculated. The electromyogram signal corresponding to the muscle excitation force was measured, and the excitation force function of an envelope curve from the electromyogram signal was extracted. The ramp input function of electrical stimulation to the carpal muscle was applied by using a device for functional electrical stimulation, and the angular displacements corresponding to steady state response were measured. Theoretical calculations of the vibration response displacements were compared with the experimental results of the angular displacements, and have shown a good agreement with the result that is linearly proportional to the excitation force magnitude. As a result, the relationship between the input current of the electrical stimulation and the excitation force magnitude was inferred. The result was shown that it can be applied to develop rehabilitation training devices.

• Journal of the Korean Society for Precision Engineering
• /
• v.22 no.3 s.168
• /
• pp.170-178
• /
• 2005

The purpose of this study is to develop a gait-event detection system, which is necessary for the cycle-to-cycle FES control of locomotion. Proposed gait event detection system consists of a signal measurement part and gait event detection part. The signal measurement was composed of the sensors and the LabVIEW program for the data acquisition and synchronization of the sensor signals. We also used a video camera and a motion capture system to get the reference gait events. Machine learning technique with ANN (artificial neural network) was adopted for automatic detection of gait events. 2 cycles of reference gait events were used as the teacher signals for ANN training and the remnants ($2\sim5$ cycles) were used fur the evaluation of the performance in gait-event detection. 14 combinations of sensor signals were used in the training and evaluation of ANN to examine the relationship between the number of sensors and the gait-event detection performance. The best combinations with minimum errors of event-detection time were 1) goniometer, foot-switch and 2) goniometer, foot-switch, accelerometer x(anterior-posterior) component. It is expected that the result of this study will be useful in the design of cycle-to-cycle FES controller.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.15 no.4
• /
• pp.817-822
• /
• 2011

This paper first proposes the speech recognition algorithm by detection of the speech and noise sections at each frame using a neural network training by back-propagation algorithm, then proposes the spectral subtraction method which removes the noises at each frame according to detection of the speech and noise sections. In this experiment, the performance of the proposed recognition system was evaluated based on the recognition rate using various speeches that are degraded by white noise and car noise. Moreover, experimental results of the noise reduction by the spectral subtraction method demonstrate using the speech and noise sections detecting by the speech recognition algorithm at each frame. Based on measuring signal-to-noise ratio, experiments confirm that the proposed algorithm is effective for the speech by corrupted the noise using signal-to-noise ratio.

• Structural Engineering and Mechanics
• /
• v.45 no.1
• /
• pp.53-67
• /
• 2013

This study aimed to develop a model to accurately predict the acceleration of structural systems during an earthquake. The acceleration and applied force of a structure were measured at current time step and the velocity and displacement were estimated through linear integration. These data were used as input to predict the structural acceleration at next time step. The computation tool used was the Volterra/Wiener neural network (VWNN) which contained the mathematical model to predict the acceleration. For alleviating problems of relatively large-dimensional and nonlinear systems, the VWNN model was utilized as the signal processing tool, including the Taylor series components in the input nodes of the neural network. The number of the intermediate layer nodes in the neural network model, containing the training and simulation stage, was evaluated and optimized. Discussions on the influences of the gradient descent with adaptive learning rate algorithm and the Levenberg-Marquardt algorithm, both for determining the network weights, on prediction errors were provided. During the simulation stage, different earthquake excitations were tested with the optimized settings acquired from the training stage to find out which of the algorithms would result in the smallest error, to determine a proper simulation model.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.11 no.12
• /
• pp.6038-6053
• /
• 2017

In this study, we propose an automatic melody extraction algorithm using deep learning. In this algorithm, feature images, generated using the energy of frequency band, are extracted from polyphonic audio files and a deep learning technique, a convolutional neural network (CNN), is applied on the feature images. In the training data, a short frame of polyphonic music is labeled as a musical note and a classifier based on CNN is learned in order to determine a pitch value of a short frame of audio signal. We want to build a novel structure of melody extraction, thus the proposed algorithm has a simple structure and instead of using various signal processing techniques for melody extraction, we use only a CNN to find a melody from a polyphonic audio. Despite of simple structure, the promising results are obtained in the experiments. Compared with state-of-the-art algorithms, the proposed algorithm did not give the best result, but comparable results were obtained and we believe they could be improved with the appropriate training data. In this paper, melody extraction and the proposed algorithm are introduced first, and the proposed algorithm is then further explained in detail. Finally, we present our experiment and the comparison of results follows.

• Speech Sciences
• /
• v.15 no.2
• /
• pp.55-65
• /
• 2008

Recently, the multi-model based speech recognizer has been used quite successfully for noisy speech recognition. For the selection of the reference HMM (hidden Markov model) which best matches the noise type and SNR (signal to noise ratio) of the input testing speech, the estimation of the SNR value using the VAD (voice activity detection) algorithm and the classification of the noise type based on the GMM (Gaussian mixture model) have been done separately in the multi-model framework. As the SNR estimation process is vulnerable to errors, we propose an efficient method which can classify simultaneously the SNR values and noise types. The KL (Kullback-Leibler) distance between the single Gaussian distributions for the noise signal during the training and testing is utilized for the classification. The recognition experiments have been done on the Aurora 2 database showing the usefulness of the model compensation method in the multi-model based speech recognizer. We could also see that further performance improvement was achievable by combining the probability density function of the MCT (multi-condition training) with that of the reference HMM compensated by the D-JA (data-driven Jacobian adaptation) in the multi-model based speech recognizer.

• Journal of Life Science
• /
• v.20 no.11
• /
• pp.1704-1710
• /
• 2010

The purpose of this study was to describe the biomechanical characteristics of stroke patients. These characteristics were obtained during walking on a Zebris system, cinematography system and EMG system. Seven female stroke patients participated in this study. The magnitude of the profiles (joint peak angle, joint peak moments, foot pressure COP, EMG data) correlated with rehabilitation training duration using t-test. The significance level selected for this study was p<0.05, t-test. Joint analysis identified significant differences in hip joint peak angle and hip joint peak moment. Foot pressure verified significant differences in gait line length of COP. The EMG signal proved significant differences in rectus femoris and vastus lateralis.

• Journal of the Institute of Convergence Signal Processing
• /
• v.5 no.3
• /
• pp.205-209
• /
• 2004

The K-means algorithm is the most widely used method among the codebook generation algorithms in vector quantization. In this paper, we propose a codebook generation algorithm using a new updating condition to enhance the codebook performance. The conventional K-means algorithm uses a fixed weight of the distance for all training iterations, but the proposed method uses different weights according to the updating condition from the new codevectors for training iterations. Then, different weights can be applied to generate codevectors at each iteration according to this condition, and it can have a similar effect to variable weights. Experimental results show that the proposed algorithm has the better codebook performance than that of K-means algorithm.