• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.20 no.3
• /
• pp.471-476
• /
• 2016

This paper presents our study on speech recognition performance prediction. Our initial study shows that a combination of speech quality measures effectively improves correlation with Word Error Rate (WER) compared to each speech measure alone. In this paper we demonstrate a new combination of various types of speech quality measures shows more significantly improves correlation with WER compared to the speech measure combination of our initial study. In our study, SNR, PESQ, acoustic model score, and MFCC distance are used as the speech quality measures. This paper also presents our speech database verification system for speech recognition employing the speech measures. We develop a WER prediction system using Gaussian mixture model and the speech quality measures as a feature vector. The experimental results show the proposed system is highly effective at predicting WER in a low SNR condition of speech babble and car noise environments.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.21 no.12
• /
• pp.2291-2297
• /
• 2017

This paper describe to extract speech measure algorithm for evaluating a speech database, and presents generating method of a speech quality measure using DNN(Deep Neural Network). In our previous study, to produce an effective speech quality measure, we propose a combination of various speech measures which are highly correlated with WER(Word Error Rate). The new combination of various types of speech quality measures in this study is more effective to predict the speech recognition performance compared to each speech measure alone. In this paper, we describe the method of extracting measure using DNN, and we change one of the combined measure from GMM(Gaussican Mixture Model) score used in the previous study to DNN score. The combination with DNN score shows a higher correlation with WER compared to the combination with GMM score.

• The Journal of the Acoustical Society of Korea
• /
• v.23 no.5
• /
• pp.383-389
• /
• 2004

In this paper. we propose the bias compensation methods and the eigenvoice method using the mean of dimensional eigenvoice to improve the performance of rapid speaker adaptation based on eigenvoice under mismatch between training and test environment. Experimental results for vocabulary-independent word recognition task (using PBW 452 DB) show that the proposed methods yield improvements for small adaptation data. We obtained about 22∼30% relative improvement by the bias compensation methods as amount of adaptation data varied from 1 to 50, and obtained 41% relative improvement in error rate by the eigenvoice method using the mean of dimensional eigenvoice with only single adaptation word.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.20 no.3
• /
• pp.464-470
• /
• 2016

This paper presents a speech recognition database verification system using speech measures, and describes a speech measure extraction algorithm which is applied to this system. In our previous study, to produce an effective speech quality measure for the system, we propose a combination of various speech measures which are highly correlated with WER (Word Error Rate). The new combination of various types of speech quality measures in this study is more effective to predict the speech recognition performance compared to each speech measure alone. In this paper, we increase the system independency by employing GMM acoustic score instead of HMM score which is obtained by a secondary speech recognition system. The combination with GMM score shows a slightly lower correlation with WER compared to the combination with HMM score, however it presents a higher relative improvement in correlation with WER, which is calculated compared to the correlation of each speech measure alone.

• Journal of the Institute of Electronics Engineers of Korea SP
• /
• v.38 no.1
• /
• pp.97-103
• /
• 2001

Hidden Markov Model (HMM) is the most widely used method in speech recognition. In general, HMM parameters are trained to have maximum likelihood (ML) for training data. Although the ML method has good performance, it dose not take account into discrimination to other words. To complement this problem, a word verification method by re-recognition of the recognized word and its similar word using the discriminative function of the two words. To find the similar word, the probability of other words to the HMM is calculated and the word showing the highest probability is selected as the similar word of the mode. To achieve discrimination to each word the weight to each state is appended to the HMM parameter. The weight is calculated by genetic algorithm. The verificator complemented discrimination of each word and reduced the error occurred by similar word. As a result of verification the total error is reduced by about 22%

• Proceedings of the Korea Information Processing Society Conference
• /
• 2005.05a
• /
• pp.767-770
• /
• 2005

네트워크상에서 전달되는 음성데이터는 전달되는 과정에서 잡음 등의 외부 요인으로 인하여 데이터에 손실이 생기는 문제가 발생한다. 이렇게 전달된 음성데이터가 음성 인식기를 통과하면 바로 음성 인식기를 통과했을 때 보다 인식률이 낮아진다. 본 연구에서는 홈 네트워크를 제어하는데 있어서 음성 인식률을 향상시키기 위해서 음성 데이터를 입력받아, 이를 음소단위 기반의 유사율 알고리즘을 적용시켜 이미 구축된 홈 네트워크 용어 관련 사전에 등록된 단어와의 유사성을 검토하여 추출된 결과로 홈 네트워크를 제어하는 방안을 제안한다. 음소단위 기반의 유사율 알고리즘과 다중발화를 이용했을 때 Threshold 값이 85% 일 경우 사전에 구축된 단어와 매칭된 인식률은 100%였으며, 사전에 없는 단어의 오인식률은 2%로 감소되었다.

• The Journal of the Acoustical Society of Korea
• /
• v.22 no.5
• /
• pp.403-407
• /
• 2003

Recently, eigenvoice approach has been widely used for rapid speaker adaptation. However, even in the eigenvoice approach, Performance improvement using very small amount of adaptation data is relatively small in comparison with that using somewhat large adaptation data because the reliable estimation of weights of eigenvoice is difficult. In this paper, we propose a rapid speaker adaptation method based on eigenvoice using the weight distribution characteristics to improve the performance on a small adaptation data. In the Experimental results on vocabulary-independent word recognition task (using PBW 452 database), the weight threshold method alleviates the problem of relatively low performance for a tiny small adaptation data. When single adaptation word is used, word error rate is reduced about 9-18% by the weight threshold method.

• Proceedings of the Korean Institute of Intelligent Systems Conference
• /
• 2002.05a
• /
• pp.1-5
• /
• 2002

The purpose of this paper is developing a speech recognition module in a wheelchair for the sake of convenience. of the disability. For this system, we used TMS320C32 as a main processor; eliminated noise by applying Winer filler while considering characteristics of noise environment in pre-processing stage, and; extracted 12 feature patterns per france using LPC&Cepstrum. Then, we implemented the hybrid form combining DTW (Dynamic Time Warping), which is generally used for isolated words in the conventional algorithms, in the recognition Part, and NN (Neural network) to prevent any error of recognition. In this research, we achieved a recognition rate of more than 96% on isolated words when DTW and Hybrid forms were individually experimented in noise environment

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.31 no.9C
• /
• pp.853-858
• /
• 2006

TMHMM(Tied Mixture Hidden Markov Model) is an important approach to reduce the number of free parameters in speech recognition. However, this model suffers from a degradation in recognition accuracy due to its GPDF (Gaussian Probability Density Function) clustering error. This paper proposes a clustering algorithm, called HCNN(Homogeneous Centroid Neural network), to cluster acoustic feature vectors in TMHMM. Moreover, the HCNN uses the heterogeneous distance measure to allocate more code vectors in the heterogeneous areas where probability densities of different states overlap each other. When applied to Korean digit isolated word recognition, the HCNN reduces the error rate by 9.39% over CNN clustering, and 14.63% over the traditional K-means clustering.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.7 no.7
• /
• pp.1380-1386
• /
• 2003

Speech recognition performance is degraded by the variation in vocal tract length among speakers. In this paper, we have used a vocal tract length normalization method wherein the frequency axis of the short-time spectrum associated with a speaker's speech is scaled to minimize the effects of speaker's vocal tract length on the speech recognition performance In order to normalize vocal tract length, we tried several frequency warping functions such as linear and piece-wise linear function. Variable interval piece-wise linear warping function is proposed to effectively model the variation of frequency axis scale due to the large variation of vocal tract length. Experimental results on TIDIGITS connected digits showed the dramatic reduction of word error rates from 2.15% to 0.53% by the proposed vocal tract normalization.