• 인터넷정보학회논문지
• /
• 제25권4호
• /
• pp.57-64
• /
• 2024

본 논문은 머신러닝을 활용하여 많은 양의 음악 데이터를 분류하여 장르 정보가 입력되어 있지 않은 음악 장르 분류 정확도 향상을 목표로 한다. 음악의 장르를 구분하기 위해 기존 연구에서 많이 사용되던 GTZAN 데이터 세트 대신 직접 데이터를 수집하고 전처리하는 방안을 제시한다. 이를 위해 GTZAN 데이터 세트보다 분류 성능이 뛰어난 데이터 세트를 생성하기 위해 Onset의 에너지 레벨이 가장 높은 일정 구간을 추출한다. 학습에 사용하는 음악 데이터의 주요 특성으로는 Mel Frequency Cepstral Coefficient(MFCC)를 포함한 57개의 특성을 이용한다. 전처리된 데이터를 통해 Support Vector Machine(SVM) 모델을 이용하여 Blues, Classical, Jazz, Country, Disco, Pop, Rock, Metal, Hiphop으로 분류한 학습 정확도가 85%를 기록하였고, 테스트 정확도가 71%를 보여주었다.

• 대한음성학회:학술대회논문집
• /
• 대한음성학회 2007년도 한국음성과학회 공동학술대회 발표논문집
• /
• pp.207-210
• /
• 2007

As the Internet is prevalent in our life, harmful contents have been increasing on the Internet, which has become a very serious problem. Among them, pornographic video is harmful as poison to our children. To prevent such an event, there are many filtering systems which are based on the keyword based methods or image based methods. The main purpose of this paper is to devise a system that classifies the pornographic videos based on the audio information. We use Mel-Cepstrum Modulation Energy (MCME) which is modulation energy calculated on the time trajectory of the Mel-Frequency cepstral coefficients (MFCC) and MFCC as the feature vector and Gaussian Mixture Model (GMM) as the classifier. With the experiments, the proposed system classified the 97.5% of pornographic data and 99.5% of non-pornographic data. We expect the proposed method can be used as a component of the more accurate classification system which uses video information and audio information simultaneously.

• 한국HCI학회:학술대회논문집
• /
• 한국HCI학회 2009년도 학술대회
• /
• pp.522-525
• /
• 2009

본 논문에서는 인터넷의 역기능으로 현대 사회에 큰 문제를 야기 시키는 음란성 유해 동영상을 내용기반으로 판별하기 위한 방법을 제안하였다. 유해 동영상에서 오디오 데이터를 이용하여 특징을 추출하였다. 사용된 오디오 특징은 주파수 스펙트럼, 자기상관, MFCC이다. 음란성의 내용이 될 수 있는 소리의 특징을 추출하였고 동영상 전체 오디오에서 해당 소리의 특징과 일치하는지를 측정하여 유해성을 판별하였다. 제안한 방법의 실험은 각 특징마다 유해 판별 측정 결과와 다중 특징을 이용한 측정 결과를 비교 수행하였다. 하나의 오디오 특징만을 추출하여 사용하였을 때 보다 다중 특징의 사용이 좋은 결과를 얻을 수 있었다.

• 전기전자학회논문지
• /
• 제17권3호
• /
• pp.241-247
• /
• 2013

This paper presents the design of a System on Programmable Chip (SoPC) based on Field Programmable Gate Array (FPGA) for speech recognition in which Mel-Frequency Cepstral Coefficients (MFCC) for speech feature extraction and Vector Quantization for recognition are used. The implementing process of the speech recognition system undergoes the following steps: feature extraction, training codebook, recognition. In the first step of feature extraction, the input voice data will be transformed into spectral components and extracted to get the main features by using MFCC algorithm. In the recognition step, the obtained spectral features from the first step will be processed and compared with the trained components. The Vector Quantization (VQ) is applied in this step. In our experiment, Altera's DE2 board with Cyclone II FPGA is used to implement the recognition system which can recognize 64 words. The execution speed of the blocks in the speech recognition system is surveyed by calculating the number of clock cycles while executing each block. The recognition accuracies are also measured in different parameters of the system. These results in execution speed and recognition accuracy could help the designer to choose the best configurations in speech recognition on SoPC.

• Journal of Information Processing Systems
• /
• 제14권4호
• /
• pp.807-823
• /
• 2018

Speaker verification system performance depends on the utterance of each speaker. To verify the speaker, important information has to be captured from the utterance. Nowadays under the constraints of limited data, speaker verification has become a challenging task. The testing and training data are in terms of few seconds in limited data. The feature vectors extracted from single frame size and rate (SFSR) analysis is not sufficient for training and testing speakers in speaker verification. This leads to poor speaker modeling during training and may not provide good decision during testing. The problem is to be resolved by increasing feature vectors of training and testing data to the same duration. For that we are using multiple frame size (MFS), multiple frame rate (MFR), and multiple frame size and rate (MFSR) analysis techniques for speaker verification under limited data condition. These analysis techniques relatively extract more feature vector during training and testing and develop improved modeling and testing for limited data. To demonstrate this we have used mel-frequency cepstral coefficients (MFCC) and linear prediction cepstral coefficients (LPCC) as feature. Gaussian mixture model (GMM) and GMM-universal background model (GMM-UBM) are used for modeling the speaker. The database used is NIST-2003. The experimental results indicate that, improved performance of MFS, MFR, and MFSR analysis radically better compared with SFSR analysis. The experimental results show that LPCC based MFSR analysis perform better compared to other analysis techniques and feature extraction techniques.

• 한국해양공학회지
• /
• 제28권6호
• /
• pp.552-559
• /
• 2014

In general, the number of underwater transient signals is very limited for research on automatic recognition. Data-dependent feature extraction is one of the most effective methods in this case. Therefore, we suggest WPCC (Wavelet packet ceptsral coefficient) as a feature extraction method. A wavelet packet best tree for each data set is formed using an entropy-based cost function. Then, every terminal node of the best trees is counted to build a common wavelet best tree. It corresponds to flexible and non-uniform filter bank reflecting characteristics for the data set. A GMM (Gaussian mixture model) is used to classify five classes of underwater transient data sets. The error rate of the WPCC is compared using MFCC (Mel-frequency ceptsral coefficients). The error rates of WPCC-db20, db40, and MFCC are 0.4%, 0%, and 0.4%, respectively, when the training data consist of six out of the nine pieces of data in each class. However, WPCC-db20 and db40 show rates of 2.98% and 1.20%, respectively, while MFCC shows a rate of 7.14% when the training data consists of only three pieces. This shows that WPCC is less sensitive to the number of training data pieces than MFCC. Thus, it could be a more appropriate method for underwater transient recognition. These results may be helpful to develop an automatic recognition system for an underwater transient signal.

• Journal of Electrical Engineering and Technology
• /
• 제12권1호
• /
• pp.373-377
• /
• 2017

This paper presents a voice-activity detection (VAD) method for sound sequences with various SNRs. For real-time VAD applications, it is inadequate to employ a post-processing for the removal of burst clippings from the VAD output decision. To tackle this problem, building on the bi-level hidden Markov model, for which a state layer is inserted into a typical hidden Markov model (HMM), we formulated a robust method for VAD not requiring any additional post-processing. In the method, a forward-inference-ratio test was devised to detect the speech endpoints and Mel-frequency cepstral coefficients (MFCC) were used as the features. Our experiment results show that, regarding different SNRs, the performance of the proposed approach is more outstanding than those of the conventional methods.

• 한국멀티미디어학회논문지
• /
• 제20권10호
• /
• pp.1637-1645
• /
• 2017

Target/Nontarget classification can be divided into the study of shape estimation of the target analysing reflected echo signal and of type classification of the target using acoustical features. In active sonar system, the feature vectors are extracted from the signal reflected from the target, and an classification algorithm is applied to determine whether the received signal is a target or not. However, received sonar signals can be distorted in the underwater environments, and the spatio-temporal characteristics of active sonar signals change according to the aspect of the target. In addition, it is very difficult to collect real sea-trial data for research. In this paper, target/non-target classification were performed using real sea-trial data. Feature vectors are extracted using MFCC(Mel-Frequency Cepstral Coefficients), filterbank energy in the Fourier spectrum and wavelet domain. For the performance verification, classification experiments were performed using backpropagation neural network classifiers.

• 한국멀티미디어학회:학술대회논문집
• /
• 한국멀티미디어학회 1998년도 추계학술발표논문집
• /
• pp.221-225
• /
• 1998

본 논문에서는 연속분포(Continuous) HMM(hidden Markov model)을 기반으로 하여 한국어 고립단어인식 시스템을 설계, 구현하였다. 시스템의 학습과 평가를 위해 자동차 항법용 음성 명령어 도메인에서 추출한 10개의 고립단어를 대상으로 음성 데이터 베이스를 구축하였다. 음성 특징 파라미터로는 MFCCs(Mel Frequency Cepstral Coefficients)와 차분(delta) MFCC 그리고 에너지(energy)를 사용하였다. 학습 데이터로부터 추출한 18개의 유사 음소(phoneme-like unit : PLU)를 인식단위로 HMM 모델을 만들었고 조음 결합 현상(채-articulation)을 모델링 하기 위해 트라이폰(triphone) 모델로 확장하였다. 인식기 평가는 학습에 참여한 음성 데이터와 학습에 참여하지 않은 화자가 발성한 음성 데이터를 이용해 수행하였으며 평균적으로 97.5%의 인식성능을 얻었다.

• 한국정보과학회:학술대회논문집
• /
• 한국정보과학회 2012년도 한국컴퓨터종합학술대회논문집 Vol.39 No.1(B)
• /
• pp.477-479
• /
• 2012

개의 발성은 성도의 물리적인 특징에 따라 고유의 특정 포먼트를 만들어 내며 개의 품종에 따라 다른 물리적 특징을 가지므로 개의 발성을 HMM(Hidden Markov Model)으로 모델링하여 개의 품종을 분류하는 연구를 하였다. 주파수 특징은 MFCC(Mel Frequency Cepstral Coefficients) 12차, 에너지 컴포넌트 1차, 델타 13차, 억셀러레이션(Acceleration) 13차, 총 39차 벡터를 사용하였다. 개의 품종 분류에 적합한 HMM 구조의 설계를 위하여 기본 좌우 모델, 좌우 모델, 좌우 모델2, 전후진 모델, 총 4가지를 제안하고 실험하여 성능을 비교분석하였다. 이 중 전후진 모델이 가장 바람직한 모델로 검증 되었다. 본 모델은 다음과 같은 장점을 갖는다. (1) 기본 좌우 모델과 마찬가지로 1~2회 발성을 갖는 데이터가 입력되어도 처음에서 마지막 상태까지의 이동단계가 최소 3번까지 가능하므로 적은 횟수의 발성 데이터도 처리가 가능하다. (2) 다수 반복된 발성 데이터의 신호도 처리가 가능하다. 즉, 본 모델은 상태의 이동이 후진도 가능하므로 5회이상 반복된 발성 데이터의 신호의 처리도 가능하다.