Pub Date : 1987-04-06DOI: 10.1109/ICASSP.1987.1169477
P. Cassereau, J. Jaffe
This paper describes the design of frequency-hopped signals for a multi-beam imaging system. A frequency hopping pattern is a frequency-coded uniform pulse train. The signal is divided into M time intervals, with each interval assigned a different frequency chosen from a set of N frequencies. A set of N patterns composed of N-1 frequencies can be generated using first-order Reed-Solomon codewords. These patterns exhibit very good correlation properties. In a frequency-hopped multi-beam imaging system, each beam is associated with a pattern and transmits a coded waveform. All N beams can be transmitted simultaneously resulting in a high scan-rate, high resolution imaging device. Furthermore, in the presence of noise and medium spreading effects, a frequency-hopped imaging device performs better than conventional systems by showing better noise rejection and less sensitivity to spreading effects.
{"title":"Frequency hopping patterns for simultaneous multiple-beam sonar imaging","authors":"P. Cassereau, J. Jaffe","doi":"10.1109/ICASSP.1987.1169477","DOIUrl":"https://doi.org/10.1109/ICASSP.1987.1169477","url":null,"abstract":"This paper describes the design of frequency-hopped signals for a multi-beam imaging system. A frequency hopping pattern is a frequency-coded uniform pulse train. The signal is divided into M time intervals, with each interval assigned a different frequency chosen from a set of N frequencies. A set of N patterns composed of N-1 frequencies can be generated using first-order Reed-Solomon codewords. These patterns exhibit very good correlation properties. In a frequency-hopped multi-beam imaging system, each beam is associated with a pattern and transmits a coded waveform. All N beams can be transmitted simultaneously resulting in a high scan-rate, high resolution imaging device. Furthermore, in the presence of noise and medium spreading effects, a frequency-hopped imaging device performs better than conventional systems by showing better noise rejection and less sensitivity to spreading effects.","PeriodicalId":140810,"journal":{"name":"ICASSP '87. IEEE International Conference on Acoustics, Speech, and Signal Processing","volume":"36 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1987-04-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133929256","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 1987-04-06DOI: 10.1109/ICASSP.1987.1169500
R. DeGroat, R. Roberts
In this paper, we discuss rank-one eigenvector updating schemes that are appropriate for tracking time-varying, narrow-band signals in noise. We show that significant reductions in computation are achieved by updating the eigenvalue decomposition (EVD) of a reduced rank version of the data covariance matrix, and that reduced rank updating yields a lower threshold breakdown than full rank updating. We also show that previously published eigenvector updating algorithms [1], [10], suffer from a linear build-up of roundoff error which becomes significant when large numbers of recursive updates are performed. We then show that exponential weighting together with pairwise Gram Schmidt partial orthogonalization at each update virtually eliminates the build-up of error making the rank-one update a useful numerical tool for recursive updating. Finally, we compare the frequency estimation performance of reduced rank weighted linear prediction and the LMS algorithm.
{"title":"An improved, highly parallel rank-one eigenvector update method with signal processing applications","authors":"R. DeGroat, R. Roberts","doi":"10.1109/ICASSP.1987.1169500","DOIUrl":"https://doi.org/10.1109/ICASSP.1987.1169500","url":null,"abstract":"In this paper, we discuss rank-one eigenvector updating schemes that are appropriate for tracking time-varying, narrow-band signals in noise. We show that significant reductions in computation are achieved by updating the eigenvalue decomposition (EVD) of a reduced rank version of the data covariance matrix, and that reduced rank updating yields a lower threshold breakdown than full rank updating. We also show that previously published eigenvector updating algorithms [1], [10], suffer from a linear build-up of roundoff error which becomes significant when large numbers of recursive updates are performed. We then show that exponential weighting together with pairwise Gram Schmidt partial orthogonalization at each update virtually eliminates the build-up of error making the rank-one update a useful numerical tool for recursive updating. Finally, we compare the frequency estimation performance of reduced rank weighted linear prediction and the LMS algorithm.","PeriodicalId":140810,"journal":{"name":"ICASSP '87. IEEE International Conference on Acoustics, Speech, and Signal Processing","volume":"21 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1987-04-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133093205","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 1987-04-06DOI: 10.1109/ICASSP.1987.1169604
Anne-Marie Derouault
One approach to large vocabulary speech recognition, is to build phonetic Markov models, and to concatenate them to obtain word models. In previous work, we already designed a recognizer based on 40 phonetic Markov machines, which accepts a 10,000 words vocabulary ([3]), and recently 200,000 words vocabulary ([5]). Since there is one machine per phoneme, these models obviously do not account for coarticulatory effects, which may lead to recognition errors. In this paper, we improve the phonetic models by using general principles about coarticulation effects on automatic phoneme recognition. We show that both the analysis of the errors made by the recognizer, and linguistic facts about phonetic context influence, suggest a method for choosing context dependent models. This method allows to limit the growing of the number of phonems, and still account for the most important coarticulation effects. We present our experiments with a system applying these principles to a set of models for French. With this new system including context-dependant machines, the phoneme recognition rate goes from 82.2% to 85.3%, and the error rate on words with a 10,000 word dictionary, is decreased from 11.2 to 9.8%.
{"title":"Context-dependent phonetic Markov models for large vocabulary speech recognition","authors":"Anne-Marie Derouault","doi":"10.1109/ICASSP.1987.1169604","DOIUrl":"https://doi.org/10.1109/ICASSP.1987.1169604","url":null,"abstract":"One approach to large vocabulary speech recognition, is to build phonetic Markov models, and to concatenate them to obtain word models. In previous work, we already designed a recognizer based on 40 phonetic Markov machines, which accepts a 10,000 words vocabulary ([3]), and recently 200,000 words vocabulary ([5]). Since there is one machine per phoneme, these models obviously do not account for coarticulatory effects, which may lead to recognition errors. In this paper, we improve the phonetic models by using general principles about coarticulation effects on automatic phoneme recognition. We show that both the analysis of the errors made by the recognizer, and linguistic facts about phonetic context influence, suggest a method for choosing context dependent models. This method allows to limit the growing of the number of phonems, and still account for the most important coarticulation effects. We present our experiments with a system applying these principles to a set of models for French. With this new system including context-dependant machines, the phoneme recognition rate goes from 82.2% to 85.3%, and the error rate on words with a 10,000 word dictionary, is decreased from 11.2 to 9.8%.","PeriodicalId":140810,"journal":{"name":"ICASSP '87. IEEE International Conference on Acoustics, Speech, and Signal Processing","volume":"58 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1987-04-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115969744","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 1987-04-06DOI: 10.1109/ICASSP.1987.1169700
Salim Roukos, M. O. Dunham
Developing accurate and robust phonetic models for the different speech sounds is a major challenge for high performance continuous speech recognition. In this paper, we introduce a new approach, called the stochastic segment model, for modelling a variable-length phonetic segment X, an L-long sequence of feature vectors. The stochastic segment model consists of 1) time-warping the variable-length segment X into a fixed-length segment Y called a resampled segment, and 2) a joint density function of the parameters of the resampled segment Y, which in this work is assumed Gaussian. In this paper, we describe the stochastic segment model, the recognition algorithm, and the iterative training algorithm for estimating segment models from continuous speech. For speaker-dependent continuous speech recognition, the segment model reduces the word error rate by one third over a hidden Markov phonetic model.
{"title":"A stochastic segment model for phoneme-based continuous speech recognition","authors":"Salim Roukos, M. O. Dunham","doi":"10.1109/ICASSP.1987.1169700","DOIUrl":"https://doi.org/10.1109/ICASSP.1987.1169700","url":null,"abstract":"Developing accurate and robust phonetic models for the different speech sounds is a major challenge for high performance continuous speech recognition. In this paper, we introduce a new approach, called the stochastic segment model, for modelling a variable-length phonetic segment X, an L-long sequence of feature vectors. The stochastic segment model consists of 1) time-warping the variable-length segment X into a fixed-length segment Y called a resampled segment, and 2) a joint density function of the parameters of the resampled segment Y, which in this work is assumed Gaussian. In this paper, we describe the stochastic segment model, the recognition algorithm, and the iterative training algorithm for estimating segment models from continuous speech. For speaker-dependent continuous speech recognition, the segment model reduces the word error rate by one third over a hidden Markov phonetic model.","PeriodicalId":140810,"journal":{"name":"ICASSP '87. IEEE International Conference on Acoustics, Speech, and Signal Processing","volume":"29 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1987-04-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115973941","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 1987-04-06DOI: 10.1109/ICASSP.1987.1169813
S. Prasad, S. Joshi
The present paper aims to present an entirely new approach for the development of "exact" recursive least squares algorithms for ARMA filtering and modeling when the inputs (asssumed here to be "white") are not observable. The approach is heavily based on the recently proposed "predictor-space" representation of ARMA processes 131 and theu se of some new, moreg eneral projection operator update formulas, breifly summarized here.
{"title":"Exact recursive least squares algorithms for ARMA modeling","authors":"S. Prasad, S. Joshi","doi":"10.1109/ICASSP.1987.1169813","DOIUrl":"https://doi.org/10.1109/ICASSP.1987.1169813","url":null,"abstract":"The present paper aims to present an entirely new approach for the development of \"exact\" recursive least squares algorithms for ARMA filtering and modeling when the inputs (asssumed here to be \"white\") are not observable. The approach is heavily based on the recently proposed \"predictor-space\" representation of ARMA processes 131 and theu se of some new, moreg eneral projection operator update formulas, breifly summarized here.","PeriodicalId":140810,"journal":{"name":"ICASSP '87. IEEE International Conference on Acoustics, Speech, and Signal Processing","volume":"29 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1987-04-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128425810","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 1987-04-06DOI: 10.1109/ICASSP.1987.1169610
J. Cioffi
There has been considerable recent interest in QR factorization for recursive solution to the least-squares adaptive-filtering problem, mainly because of the good numerical properties of QR factorizations. Early work by Gentleman and Kung (1981) and McWhirter (1983) has produced triangular systolic arrays of N2/2 processors that solve the Recursive Least Squares (RLS) adaptive-filtering problem (where N is the size of the adaptive filter). Here, we introduce a more computationally efficient solution to the QR RLS problem that requires only O(N) computations per time update, when the input has the usual shift-invariant property. Thus, computation and implementation requirements are reduced by an order of magnitude. The new algorithms are based on a structure that is neither a transversal filter nor a lattice, but can be best characterized by a functionally equivalent set of parameters that represent the time-varying "least-squares frequency transforms" of the input sequences. Numerical stability can be insured by implementing computations as 2 × 2 orthogonal (Givens) rotations.
{"title":"A fast QR/Frequency-domain RLS adaptive filter","authors":"J. Cioffi","doi":"10.1109/ICASSP.1987.1169610","DOIUrl":"https://doi.org/10.1109/ICASSP.1987.1169610","url":null,"abstract":"There has been considerable recent interest in QR factorization for recursive solution to the least-squares adaptive-filtering problem, mainly because of the good numerical properties of QR factorizations. Early work by Gentleman and Kung (1981) and McWhirter (1983) has produced triangular systolic arrays of N2/2 processors that solve the Recursive Least Squares (RLS) adaptive-filtering problem (where N is the size of the adaptive filter). Here, we introduce a more computationally efficient solution to the QR RLS problem that requires only O(N) computations per time update, when the input has the usual shift-invariant property. Thus, computation and implementation requirements are reduced by an order of magnitude. The new algorithms are based on a structure that is neither a transversal filter nor a lattice, but can be best characterized by a functionally equivalent set of parameters that represent the time-varying \"least-squares frequency transforms\" of the input sequences. Numerical stability can be insured by implementing computations as 2 × 2 orthogonal (Givens) rotations.","PeriodicalId":140810,"journal":{"name":"ICASSP '87. IEEE International Conference on Acoustics, Speech, and Signal Processing","volume":"6 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1987-04-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131286451","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 1987-04-06DOI: 10.1109/ICASSP.1987.1169330
R. Rastogi, P. Gupta, R. Kumaresan
Estimation of angles of arrival of plane waves from data observed at an array of sensors is performed with a network of interconnected, instantaneous, saturating non-linear elements called neurons. The networks use the observed data to decide which among a large number of hypothesized angles of arrivals best fits the data. A possible stochastic-digital implementation of such a network is also indicated.
{"title":"Array signal processing with interconnected Neuron-like elements","authors":"R. Rastogi, P. Gupta, R. Kumaresan","doi":"10.1109/ICASSP.1987.1169330","DOIUrl":"https://doi.org/10.1109/ICASSP.1987.1169330","url":null,"abstract":"Estimation of angles of arrival of plane waves from data observed at an array of sensors is performed with a network of interconnected, instantaneous, saturating non-linear elements called neurons. The networks use the observed data to decide which among a large number of hypothesized angles of arrivals best fits the data. A possible stochastic-digital implementation of such a network is also indicated.","PeriodicalId":140810,"journal":{"name":"ICASSP '87. IEEE International Conference on Acoustics, Speech, and Signal Processing","volume":"241 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1987-04-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114464633","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 1987-04-06DOI: 10.1109/ICASSP.1987.1169476
F. Itakura, T. Umezaki
We present a novel spectral distance measure based on the smoothed LPC group delay spectrum which gives a stable recognition performance under variable frequency transfer characteristics and additive noise. The weight of the n-th cepstral coefficients in our measure is given byW_{n} = n^{s}. exp(-n^{2}/2tau^{2})which can be adjusted by selecting proper values ofsand τ. In order to optimize the parameters of this distance measure, extensive experiments are carried out in a speaker-dependent isolated word recognition system using a standard dynamic time warping technique. The input speech data used here is a set of phonetically very similar 68 Japanese city name pairs spoken by male speakers. The experimental results show that our distance measure gives a robust recognition rate in spite of the variation in frequency characteristics and signal to noise ratio(SNR). In noisy situations of segmental SNR 20 dB, the recognition rate was more than 13% higher than that obtained by using the standard Euclidean cepstral distance measure. Finally, it is shown that the optimum value ofsis approximately 1, and the optimum range of τΔT is about 1 ms.
提出了一种基于平滑LPC群延迟谱的频谱距离测量方法，该方法在可变频率传输特性和加性噪声条件下具有稳定的识别性能。在我们的测量中，第n个倒谱系数的权重由{w＿n} = n^{s}给出。exp (-n^2{/}2 tau ^2{)}，可以通过选择适当的sand τ值来调整。为了优化这种距离度量的参数，我们在一个依赖于说话人的孤立词识别系统中使用标准的动态时间规整技术进行了大量的实验。这里使用的输入语音数据是一组语音非常相似的68个日本城市名称对，由男性说话者说出。实验结果表明，在频率特性和信噪比变化的情况下，我们的距离测量方法具有良好的鲁棒识别率。在信噪比为20 dB的噪声情况下，识别率大于13% higher than that obtained by using the standard Euclidean cepstral distance measure. Finally, it is shown that the optimum value ofsis approximately 1, and the optimum range of τΔT is about 1 ms.
{"title":"Distance measure for speech recognition based on the smoothed group delay spectrum","authors":"F. Itakura, T. Umezaki","doi":"10.1109/ICASSP.1987.1169476","DOIUrl":"https://doi.org/10.1109/ICASSP.1987.1169476","url":null,"abstract":"We present a novel spectral distance measure based on the smoothed LPC group delay spectrum which gives a stable recognition performance under variable frequency transfer characteristics and additive noise. The weight of the n-th cepstral coefficients in our measure is given byW_{n} = n^{s}. exp(-n^{2}/2tau^{2})which can be adjusted by selecting proper values ofsand τ. In order to optimize the parameters of this distance measure, extensive experiments are carried out in a speaker-dependent isolated word recognition system using a standard dynamic time warping technique. The input speech data used here is a set of phonetically very similar 68 Japanese city name pairs spoken by male speakers. The experimental results show that our distance measure gives a robust recognition rate in spite of the variation in frequency characteristics and signal to noise ratio(SNR). In noisy situations of segmental SNR 20 dB, the recognition rate was more than 13% higher than that obtained by using the standard Euclidean cepstral distance measure. Finally, it is shown that the optimum value ofsis approximately 1, and the optimum range of τΔT is about 1 ms.","PeriodicalId":140810,"journal":{"name":"ICASSP '87. IEEE International Conference on Acoustics, Speech, and Signal Processing","volume":"12 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1987-04-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129386887","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 1987-04-06DOI: 10.1109/ICASSP.1987.1169487
K. Yu
This paper is concerned with mixed time-frequency signal processing using Wigner Distribution Function (WDF). This approach is based upon the generation of the mixed time-frequency representation (MTFR) of a signal, processing of that representation in the mixed time-frequency domain, and obtaining a filtered output by an inverse operation or approximation procedure. Various signal processing operations are formulated. Validity condition for the resulting MTFR are investigated. An inverse operation can be applied as an exact procedure if the output MTFR is a valid WDF, it can be regarded as an approximation procedure if the resulting MTFR is not admissible. Other approximations can be formulated. One scheme depends on the projection of the MTFR onto the space of valid WDF. The other scheme depends on the modification of the filtering function in a minimal way such that the resulting MTFR is valid.
{"title":"Signal representation and processing in the mixed time-frequency domain","authors":"K. Yu","doi":"10.1109/ICASSP.1987.1169487","DOIUrl":"https://doi.org/10.1109/ICASSP.1987.1169487","url":null,"abstract":"This paper is concerned with mixed time-frequency signal processing using Wigner Distribution Function (WDF). This approach is based upon the generation of the mixed time-frequency representation (MTFR) of a signal, processing of that representation in the mixed time-frequency domain, and obtaining a filtered output by an inverse operation or approximation procedure. Various signal processing operations are formulated. Validity condition for the resulting MTFR are investigated. An inverse operation can be applied as an exact procedure if the output MTFR is a valid WDF, it can be regarded as an approximation procedure if the resulting MTFR is not admissible. Other approximations can be formulated. One scheme depends on the projection of the MTFR onto the space of valid WDF. The other scheme depends on the modification of the filtering function in a minimal way such that the resulting MTFR is valid.","PeriodicalId":140810,"journal":{"name":"ICASSP '87. IEEE International Conference on Acoustics, Speech, and Signal Processing","volume":"24 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1987-04-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127784533","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 1987-04-06DOI: 10.1109/ICASSP.1987.1169572
S. Parthasarathy, D. Tufts
The analysis of signals that can be represented as a linear combination of exponentially damped sinusoids where the values of damping factors, frequencies, and the linear combination coefficients change at certain transition times is considered. These transitions represent the opening and closing of the glottis in the case of speech signals. Techniques are presented for the accurate estimation of the exponential parameters and the times of transition, from noise corrupted observations of the signal. The exponential parameters are obtained by improved linear prediction techniques using low-rank approximations, and further refined by an iterative least-squares technique with stability constraints imposed on the damping factors. Optimal estimates (in the least-squares sense) of the time of transition are presented. Our knowledge of the signal structure is used to obtain improved performance and also a computationally efficient estimation algorithm. Experiments with real, connected speech indicate that the speech waveforms can be accurately represented from a small number of parameters using the analysis presented here.
{"title":"Signal modeling by exponential segments and application in voiced speech analysis","authors":"S. Parthasarathy, D. Tufts","doi":"10.1109/ICASSP.1987.1169572","DOIUrl":"https://doi.org/10.1109/ICASSP.1987.1169572","url":null,"abstract":"The analysis of signals that can be represented as a linear combination of exponentially damped sinusoids where the values of damping factors, frequencies, and the linear combination coefficients change at certain transition times is considered. These transitions represent the opening and closing of the glottis in the case of speech signals. Techniques are presented for the accurate estimation of the exponential parameters and the times of transition, from noise corrupted observations of the signal. The exponential parameters are obtained by improved linear prediction techniques using low-rank approximations, and further refined by an iterative least-squares technique with stability constraints imposed on the damping factors. Optimal estimates (in the least-squares sense) of the time of transition are presented. Our knowledge of the signal structure is used to obtain improved performance and also a computationally efficient estimation algorithm. Experiments with real, connected speech indicate that the speech waveforms can be accurately represented from a small number of parameters using the analysis presented here.","PeriodicalId":140810,"journal":{"name":"ICASSP '87. IEEE International Conference on Acoustics, Speech, and Signal Processing","volume":"23 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1987-04-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"134059338","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：