• The Journal of the Acoustical Society of Korea
• /
• v.29 no.2E
• /
• pp.86-99
• /
• 2010

This paper analyzes the performance of various single channel speech enhancement algorithms when they are applied to automatic speech recognition (ASR) systems as a preprocessor. The functional modules of speech enhancement systems are first divided into four major modules such as a gain estimator, a noise power spectrum estimator, a priori signal to noise ratio (SNR) estimator, and a speech absence probability (SAP) estimator. We investigate the relationship between speech recognition accuracy and the roles of each module. Simulation results show that the Wiener filter outperforms other gain functions such as minimum mean square error-short time spectral amplitude (MMSE-STSA) and minimum mean square error-log spectral amplitude (MMSE-LSA) estimators when a perfect noise estimator is applied. When the performance of the noise estimator degrades, however, MMSE methods including the decision directed module to estimate a priori SNR and the SAP estimation module helps to improve the performance of the enhancement algorithm for speech recognition systems.

• Journal of the Korea Institute of Military Science and Technology
• /
• v.25 no.3
• /
• pp.219-230
• /
• 2022

In implementing a robust automatic target recognition(ATR) system with synthetic aperture radar(SAR) imagery, one of the most important issues is accurate classification of target variants, which are the same targets with different serial numbers, configurations and versions, etc. In this paper, a deep learning network with channel attention modules is proposed to cope with the recognition problem for target variants based on the previous research findings that the channel attention mechanism selectively emphasizes the useful features for target recognition. Different from other existing attention methods, this paper employs the channel attention modules without dimensionality reduction along the channel direction from which direct correspondence between feature map channels can be preserved and the features valuable for recognizing SAR target variants can be effectively derived. Experiments with the public benchmark dataset demonstrate that the proposed scheme is superior to the network with other existing channel attention modules.

• Phonetics and Speech Sciences
• /
• v.8 no.2
• /
• pp.31-39
• /
• 2016

In emergency dispatching at 119 Command & Dispatch Center, some inconsistencies between the 'standard emergency aid system' and 'dispatch protocol,' which are both mandatory to follow, cause inefficiency in the dispatcher's performance. If an emergency dispatch system uses automatic speech recognition (ASR) to process the dispatcher's protocol speech during the case registration, it instantly extracts and provides the required information specified in the 'standard emergency aid system,' making the rescue command more efficient. For this purpose, we have developed a Korean large vocabulary continuous speech recognition system for 400,000 words to be used for the emergency dispatch system. The 400,000 words include vocabulary from news, SNS, blogs and emergency rescue domains. Acoustic model is constructed by using 1,300 hours of telephone call (8 kHz) speech, whereas language model is constructed by using 13 GB text corpus. From the transcribed corpus of 6,600 real telephone calls, call logs with emergency rescue command class and identified major symptom are extracted in connection with the rescue activity log and National Emergency Department Information System (NEDIS). ASR is applied to emergency dispatcher's repetition utterances about the patient information. Based on the Levenshtein distance between the ASR result and the template information, the emergency patient information is extracted. Experimental results show that 9.15% Word Error Rate of the speech recognition performance and 95.8% of emergency response detection performance are obtained for the emergency dispatch system.

• KIPS Transactions on Software and Data Engineering
• /
• v.5 no.9
• /
• pp.419-424
• /
• 2016

Named entity recognition is required to improve the retrieval accuracy of patent documents or similar patents in the claims and patent descriptions. In this paper, we proposed an automatic named entity recognition for patents by using a conditional random field that is one of the best methods in machine learning research. Named entity recognition system has been constructed from the training set of tagged corpus with 660,000 words and 70,000 words are used as a test set for evaluation. The experiment shows that the accuracy is 93.6% and the Kappa coefficient is 0.67 between manual tagging and automatic tagging system. This figure is better than the Kappa coefficient 0.6 for manually tagged results and it shows that automatic named entity tagging system can be used as a practical tagging for patent documents in replacement of a manual tagging.

• The Journal of the Acoustical Society of Korea
• /
• v.18 no.3
• /
• pp.68-72
• /
• 1999

The choice of parameters from various lip information and the robustness of extracting lip parameters play important roles in the performance of bimodal speech recognition system. In this paper, lip parameters are extracted by using an automatic extraction algorithm and inner lip parameters effect on the recognition rate more than outer lip parameters. Compared with a manual extraction algorithm, the automatic extraction method is evaluated about its robustness.

• Journal of the Korean Institute of Telematics and Electronics B
• /
• v.31B no.4
• /
• pp.77-87
• /
• 1994

This paper presents recognition of raised alphanumeric markings on rubber tires for their automatic classification. Raised alphanumeric markings on rubber tires have different characteristics as compared to those of printed characters. In the preprocessing step, we first determine the rotation angle using the Hough transform and align markings, then separate each character using vertical and horizontal projections. In the recognition step, we use several features such as width of a character, cross point, partial projection, and distance feature to recognize characters hierarchically. The computer simulation result shows that the proposed system can be successfully applied to the industrial automation of rubber tires classification.

• The Journal of the Acoustical Society of Korea
• /
• v.17 no.4E
• /
• pp.11-16
• /
• 1998

For large vocabulary speech recognition using HMMs, context-dependent subword units have been often employed. However, when context-dependent phone models are used, they result in a system which has too may parameters to train. The problem of too many parameters and too little training data is absolutely crucial in the design of a statistical speech recognizer. Furthermore, when building large vocabulary speech recognition systems, unseen triphone problem is unavoidable. In this paper, we propose the modified phonetic decision tree algorithm for the automatic prediction of unseen triphones which has advantages solving these problems through following two experiments in Japanese contexts. The baseline experimental results show that the modified tree based clustering algorithm is effective for clustering and reducing the number of states without any degradation in performance. The task experimental results show that our proposed algorithm also has the advantage of providing a automatic prediction of unseen triphones.

• International Journal of Fuzzy Logic and Intelligent Systems
• /
• v.4 no.2
• /
• pp.176-181
• /
• 2004

This paper introduces a Web site login system that uses user's native voice to improve the bother of remembering the ID and password in order to login the Web site. The DTW method that applies fuzzy inference is used as the speaker recognition algorithm. We get the ACC(Average Cepstrum Coefficient) membership function by each degree, by using the LPC that models the vocal chords, to block the recorded voice that is problem for the speaker recognition. We infer the existence of the recorded voice by setting on the basis of the number of zeros that is the value of the ACC membership function, and on the basis of the average value of the ACC membership function. We experiment the six Web sites for the six subjects and get the result that protects the recorded voice about 98% that is recorded by the digital recorder.

• Phonetics and Speech Sciences
• /
• v.8 no.2
• /
• pp.23-30
• /
• 2016

This paper describes some implementation schemes of CNN in view of mini-batch DNN training for efficient second order optimization. This uses same procedure updating parameters of DNN to train parameters of CNN by simply arranging an input image as a sequence of local patches, which is actually equivalent with mini-batch DNN training. Through this conversion, second order optimization providing higher performance can be simply conducted to train the parameters of CNN. In both results of image recognition on MNIST DB and syllable automatic speech recognition, our proposed scheme for CNN implementation shows better performance than one based on DNN.

• Journal of the Korean Institute of Telematics and Electronics B
• /
• v.32B no.11
• /
• pp.1455-1463
• /
• 1995

A new automatic tracking & recognition algorithm which decides the opening & the closing states of subject's eye and isn't affected by the subject's background is proposed. And it was tested in circumstances in which subject's background was not restricted using the developed system, ATRS(Automatic Tracking & Recognition System). The significant characteristic of the ATRS is new movement detection of object that is a body in motion with accelated velocity and it doesn't need any extra hardware except a formal CCD camera and an image grabber but it works so well and so fast. The ATRS would be particularly well suited to a way of communications of patients in a hospital, who can not communicate otherwise.