• Proceedings of the IEEK Conference
• /
• 2003.07e
• /
• pp.2164-2167
• /
• 2003

In this paper we propose effective speech recognizer through recognition experiments for three feature parameters(PCA, ICA and MFCC) using SVM(Support Vector Machine) classifier In general, SVM is classification method which classify two class set by finding voluntary nonlinear boundary in vector space and possesses high classification performance under few training data number. In this paper we compare recognition result for each feature parameter and propose ICA feature as the most effective parameter

• Journal of Korea Multimedia Society
• /
• v.19 no.2
• /
• pp.128-137
• /
• 2016

In mobile communication, air conduction(AC) speech signal had been commonly used, but it was easily affected by ambient noise environment such as emergency, military action and rescue. To overcome the weakness of the AC speech signal, bone conduction(BC) speech signal have been used. The BC speech signal is transmitted through bone vibration, so it is affected less by the background noise. In this paper, we proposed noise cancellation algorithm of the BC speech signal using noise feature of decomposed bands. The proposed algorithm consist of three steps. First, the BC speech signal is divided into 17 bands using perceptual wavelet packet decomposition. Second, threshold is calculated by noise feature during short time of separated-band and compared to absolute average of the signal frame. Therefore, the speech and noise parts are detected. Last, the detected noise parts are removed and then, noise eliminated bands are re-synthesised. In order to confirm the efficiency of the proposed algorithm, we compared the proposed algorithm with conventional algorithm. And the proposed algorithm has better performance than the conventional algorithm.

• The Journal of the Acoustical Society of Korea
• /
• v.19 no.2
• /
• pp.40-45
• /
• 2000

In speech corpus generation and speech recognition, it is sometimes needed to segment the input speech data without any prior knowledge. A method to accomplish this kind of segmentation, often called as blind segmentation, or acoustic segmentation, is to find boundaries which minimize the Euclidean distances among the feature vectors of each segments. However, the use of this metric alone is prone to errors because of the fluctuations or variations of the feature vectors within a segment. In this paper, we introduce the principal component analysis method to take the trend of feature vectors into consideration, so that the proposed distance measure be the distance between feature vectors and their projected points on the principal components. The proposed distance measure is applied in the LBDP(level building dynamic programming) algorithm for an experimentation of continuous speech segmentation. The result was rather promising, resulting in 3-6% reduction in deletion rate compared to the pure Euclidean measure.

• MALSORI
• /
• no.63
• /
• pp.125-137
• /
• 2007

Recently, many techniques have been proposed to improve the noise robustness for speaker verification. In this paper, we consider the feature recombination technique in multi-band approach. In the conventional feature recombination for speaker verification, to compute the likelihoods of speaker models or universal background model, whole feature components are used. This computation method is not effective in a view point of multi-band approach. To deal with non-effectiveness of the conventional feature recombination technique, we introduce a subband likelihood computation, and propose a modified feature recombination using subband likelihoods. In decision step of speaker verification system in noise environments, a few very low likelihood scores of a speaker model or universal background model cause speaker verification system to make wrong decision. To overcome this problem, a reliable feature selection method is proposed. The low likelihood scores of unreliable feature are substituted by likelihood scores of the adaptive noise model. In here, this adaptive noise model is estimated by maximum a posteriori adaptation technique using noise features directly obtained from noisy test speech. The proposed method using subband-based reliable feature selection obtains better performance than conventional feature recombination system. The error reduction rate is more than 31 % compared with the feature recombination-based speaker verification system.

• ETRI Journal
• /
• v.46 no.1
• /
• pp.22-34
• /
• 2024

Exposure to varied noisy environments impairs the recognition performance of artificial intelligence-based speech recognition technologies. Degraded-performance services can be utilized as limited systems that assure good performance in certain environments, but impair the general quality of speech recognition services. This study introduces an audiovisual speech recognition (AVSR) model robust to various noise settings, mimicking human dialogue recognition elements. The model converts word embeddings and log-Mel spectrograms into feature vectors for audio recognition. A dense spatial-temporal convolutional neural network model extracts features from log-Mel spectrograms, transformed for visual-based recognition. This approach exhibits improved aural and visual recognition capabilities. We assess the signal-to-noise ratio in nine synthesized noise environments, with the proposed model exhibiting lower average error rates. The error rate for the AVSR model using a three-feature multi-fusion method is 1.711%, compared to the general 3.939% rate. This model is applicable in noise-affected environments owing to its enhanced stability and recognition rate.

• 전자공학회논문지 IE
• /
• v.43 no.2
• /
• pp.22-29
• /
• 2006

Recently the use of whispered speech has increased due to mobile phone and the necessity of whispered speech recognition is increasing. So various feature vectors, which are mainly used for speech recognition, are applied to their HMMs, normal speech models, whispered speech models, and integrated models with normal speech and whispered speech so as to find out suitable recognition system for whispered speech. The experimental results of recognition test show that the recognition rate of whispered speech applied to normal speech models is too low to be used in practical applications, but separate whispered speech models recognize whispered speech with the highest rates at least 85%. And also integrated models with normal speech and whispered speech score acceptable recognition rate but more study is needed to increase recognition rate. MFCE and PLCC feature vectors score higher recognition rate when applied to separate whispered speech models, but PLCC is the best when a lied to integrated models with normal speech and whispered speech.

• Proceedings of the IEEK Conference
• /
• 2004.08c
• /
• pp.764-767
• /
• 2004

The performance of a speech recognition system is generally degraded in telephone environment because of distortions caused by background noise and various channel characteristics. In this paper, data-driven temporal filters are investigated to improve the performance of a specific recognition task such as telephone speech. Three different temporal filtering methods are presented with recognition results for Korean connected-digit telephone speech. Filter coefficients are derived from the cepstral domain feature vectors using the principal component analysis.

• Journal of Institute of Control, Robotics and Systems
• /
• v.10 no.9
• /
• pp.763-770
• /
• 2004

The important issue of emotion recognition from speech is a feature extracting and pattern classification. Features should involve essential information for classifying the emotions. Feature selection is needed to decompose the components of speech and analyze the relation between features and emotions. Specially, a pitch of speech components includes much information for emotion. Accordingly, this paper searches the relation of emotion to features such as the sound loudness, pitch, etc. and classifies the emotions by using the statistic of the collecting data. This paper deals with the method of recognizing emotion from the sound. The most important emotional component of sound is a tone. Also, the inference ability of a brain takes part in the emotion recognition. This paper finds empirically the emotional components from the speech and experiment on the emotion recognition. This paper also proposes the recognition method using these emotional components and the transition probability.

• Proceedings of the KSPS conference
• /
• 2003.10a
• /
• pp.149-152
• /
• 2003

In this paper, data-driven temporal filter methods[1] are investigated for robust feature extraction. A principal component analysis technique is applied to the time trajectories of feature sequences of training speech data to get appropriate temporal filters. We did recognition experiments on the Korean connected digit telephone speech database released by SITEC, with data-driven temporal filters. Experimental results are discussed with our findings.

• Speech Sciences
• /
• v.9 no.4
• /
• pp.253-263
• /
• 2002

In this paper, we propose new algorithm of speaker recognition which identifies the speaker using the information obtained by the intensive speech feature analysis such as pitch, intensity, duration, and formant, which are crucial parameters of individual voice, for candidates of high percentage of wrong recognition in the existing speaker recognition algorithm. For testing the power of discrimination of individual parameter, DTW (Dynamic Time Warping) is used. We newly set the range of threshold which affects the power of discrimination in speech verification such that the candidates in the new range of threshold are finally discriminated in the next stage of sound parameter analysis. In the speaker verification test by using voice DB which consists of secret words of 25 males and 25 females of 8 kHz 16 bit, the algorithm we propose shows about 1% of performance improvement to the existing algorithm.