• 대한음성학회지:말소리
• /
• 제67호
• /
• pp.181-193
• /
• 2008

In this paper, we propose a vocabulary coverage improvement method for embedded continuous speech recognition (CSR) using a part-of-speech (POS) tagged corpus. We investigate 152 POS tags defined in Lancaster-Oslo-Bergen (LOB) corpus and word-POS tag pairs. We derive a new vocabulary through word addition. Words paired with some POS tags have to be included in vocabularies with any size, but the vocabulary inclusion of words paired with other POS tags varies based on the target size of vocabulary. The 152 POS tags are categorized according to whether the word addition is dependent of the size of the vocabulary. Using expert knowledge, we classify POS tags first, and then apply different ways of word addition based on the POS tags paired with the words. The performance of the proposed method is measured in terms of coverage and is compared with those of vocabularies with the same size (5,000 words) derived from frequency lists. The coverage of the proposed method is measured as 95.18% for the test short message service (SMS) text corpus, while those of the conventional vocabularies cover only 93.19% and 91.82% of words appeared in the same SMS text corpus.

• 음성과학
• /
• 제1권
• /
• pp.295-314
• /
• 1997

In this study, we will implement a flexible formant analysis and synthesis system. In the analysis part, the two-channel (i.e., speech & EGG signals) approach is investigated for accurate estimation of formant information. The EGG signal is used for extracting exact pitch information that is needed for the pitch synchronous LPC analysis and closed phase LPC analysis. In the synthesis part, Klatt formant synthesizer is modified so that the user can change synthesis parameters arbitarily. Experimental results demonstrate the superiority of the two-channel analysis method over the one-channel(speech signal only) method in analysis as well as in synthesis. The implemented system is expected to be very helpful for studing the effects of synthesis parameters on the quality of synthetic speech and for the development of Korean text-to-speech(TTS) system with the formant synthesis method.

• 대한음성학회:학술대회논문집
• /
• 대한음성학회 2003년도 10월 학술대회지
• /
• pp.25-28
• /
• 2003

The aim of this paper is to analyze the pathological voice by separating signal into periodic and aperiodic part. Separation was peformed recursively from the residual signal of voice signal. Based on initial estimation of aperiodic part of spectrum, aperiodic part is decided from the extrapolation method. Periodic part is decided by subtracting aperiodic part from the original spectrum. A parameter HNR is derived based on the separation. Parameter value statistics are compared with those of Jitter and Shimmer for normal, benign and malignant cases.

• 한국소음진동공학회:학술대회논문집
• /
• 한국소음진동공학회 2005년도 춘계학술대회논문집
• /
• pp.334-337
• /
• 2005

The speech intelligibility is one of the most important factors in communication system. The speech intelligibility is related with speech to noise ratio. To enhance the speech to noise ratio, background noise reduction techniques are being developed. As a part of solution to noise reduction, this paper introduces directional microphone using beamforming method and speech filtering method. The directional microphone narrows the spatial range of processing signal into the direction of the target speech signal. The noise signal located in the same direction with speech still remains in the processing signal. To sort this mixed signal into speech and noise, as a following step, a speech-filtering method is applied to pick up only the speech signal from the processed signal. The speech filtering method is based on the characteristics of speech signal itself. The combined directional microphone and speech filtering method gives enhanced performance to speech intelligibility in communication system.

• 대한전자공학회:학술대회논문집
• /
• 대한전자공학회 2009년도 정보 및 제어 심포지움 논문집
• /
• pp.40-42
• /
• 2009

This paper presents an algorithm, which is based on Otsu's method, for accurate and robust endpoint detection for speech recognition under noisy environments. The features are extracted in time domain, and then an optimal threshold is selected by minimizing the discriminant criterion, so as to maximize the separability of the speech part and environment part. The simulation results show that the method play a good performance in detection accuracy.

• The Journal of the Acoustical Society of Korea
• /
• 제22권2E호
• /
• pp.62-68
• /
• 2003

This paper addresses a novel noise-compensation scheme to solve the mismatch problem between training and testing condition for the automatic speech recognition (ASR) system, specifically in car environment. The conventional spectral subtraction schemes rely on the signal-to-noise ratio (SNR) such that attenuation is imposed on that part of the spectrum that appears to have low SNR, and accentuation is made on that part of high SNR. However, these schemes are based on the postulation that the power spectrum of noise is in general at the lower level in magnitude than that of speech. Therefore, while such postulation is adequate for high SNR environment, it is grossly inadequate for low SNR scenarios such as that of car environment. This paper proposes an efficient spectral subtraction scheme focused specifically to low SNR noisy environment by extracting harmonics distinctively in speech spectrum. Representative experiments confirm the superior performance of the proposed method over conventional methods. The experiments are conducted using car noise-corrupted utterances of Aurora2 corpus.

• 말소리와 음성과학
• /
• 제8권4호
• /
• pp.49-61
• /
• 2016

The present study investigates the perception of Korean phonation types by native speakers of Aymara. Perception tests were conducted on two sets of Korean speech materials to determine correspondence between Korean and Aymara 3-way contrasts and to find out which of the consonantal and vocalic part of the syllable is more influential in the perception of Korean phonation types. A set of manipulated stimuli, as well as a set of 12 spontaneous words, were prepared for the tests. The first syllable of the 12 Korean bisyllabic words of 3 series of phonation types(Lenis, Aspirated, and Fortis) in 4 places of articulation were split into consonantal and vocalic parts. And then the two parts were combined to form 9 tokens of CV sequences respectively for each place of articulation. Native speakers of Aymara were forced to match Korean stimuli with one of the 15 Aymara words which represent 3 series of consonant types(plain, aspirated, and ejective) in 5 places of articulation(bilabial, alveolar, palatal, velar, and uvular). Results showed that the consonantal part is more influential than the vocalic part to the Aymara subjects' perception of Korean phonation types when the consonantal part is Aspirated in its phonation type, but the vocalic part is more influential than the consonantal part when the consonantal part is Lenis or Fortis in its phonation type. Response analysis showed that Aymara subjects tend to match Korean stops to Aymara ones in such a way that Lenis corresponds to aspirated, Aspirated to aspirated, and Fortis to plain.

• 인지과학
• /
• 제22권3호
• /
• pp.327-345
• /
• 2011

본 논문에서는 형태소 분석기를 사용하지 않는 음절 단위의 한국어 품사 태깅 방법론을 제안한다. 기존 연구에서 한국어 품사 태거는 형태소 분석기가 생성한 결과 중에서 문맥에 가장 잘 맞는 형태소/품사 열을 결정하는 데 반하여, 본 논문에서 제안한 방법론에서는 품사열을 결정할 뿐만 아니라 형태소도 생성한다. 398,632 어절의 학습 데이터로 학습을 하고 33,467 어절의 평가 데이터로 성능 평가를 한 결과 어절 단위의 정확도가 96.31%인 것으로 나타났다.

• 제어로봇시스템학회:학술대회논문집
• /
• 제어로봇시스템학회 2003년도 ICCAS
• /
• pp.1042-1045
• /
• 2003

In this paper, we propose the speaker identification system that uses vowel that has speaker's characteristic. System is divided to speech feature extraction part and speaker identification part. Speech feature extraction part extracts speaker's feature. Voiced speech has the characteristic that divides speakers. For vowel extraction, formants are used in voiced speech through frequency analysis. Vowel-a that different formants is extracted in text. Pitch, formant, intensity, log area ratio, LP coefficients, cepstral coefficients are used by method to draw characteristic. The cpestral coefficients that show the best performance in speaker identification among several methods are used. Speaker identification part distinguishes speaker using Neural Network. 12 order cepstral coefficients are used learning input data. Neural Network's structure is MLP and learning algorithm is BP (Backpropagation). Hidden nodes and output nodes are incremented. The nodes in the incremental learning neural network are interconnected via weighted links and each node in a layer is generally connected to each node in the succeeding layer leaving the output node to provide output for the network. Though the vowel extract and incremental learning, the proposed system uses low learning data and reduces learning time and improves identification rate.

• 한국콘텐츠학회논문지
• /
• 제9권2호
• /
• pp.18-26
• /
• 2009

규칙 정보와 통계 정보를 이용하는 복합적 품사 태깅은 통계를 기반으로 하는 방법의 견고함과 확장성을 가지고, 통계 정보에 벗어나는 언어현상들을 규칙 정보를 이용하여 해결함으로서 높은 정확도를 가질 수 있다. 하지만 기존의 연구는 규칙 정보의 제한적인 적용범위 때문에 통계 정보에 벗어나는 언어 현상을 처리할 수 없는 경우가 발생하게 된다. 본 논문에서는 이를 해결하기 위하여 어휘의 사전적 의미와 문맥적 관계를 반영할 수 있는 "어휘별 중의성 제거 규칙"을 제안한다. 어휘별 중의성 제거 규칙은 세종 말뭉치로 부터 말뭉치 데이터를 형태소 분석하여 상위 50%의 중의성 어휘에 대한 사전적 의미와 문맥적 관계를 고려한 품사 태깅 정보를 추출하고 이것을 규칙으로 만든 것이며, 현재까지 총 1,815개로 구성되어 있다. 어휘별 중의성 제거 규칙을 기존의 복합적 품사 태깅 시스템에 적용하여 품사 태깅의 정확도를 높일 수 있었다.