Signal conditioning techniques for robust speech recognition

Rahim, Mazin G.; Juang, Biing‐Hwang; Chou, Wu; Buhrke, E.R.

doi:10.1109/97.489062

Cited by 49 publications

(29 citation statements)

References 4 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Given that each x i can be understood as a varying excitation convolved by a channel and the cepstrum transforms a convolutional mix into an addition, the mean value will be the unvarying part, which corresponds to the channel. This idea has been used in speech processing to determine the spectral characteristics of channels [28].…”

Section: ) Cepstral Analysismentioning

confidence: 99%

Type 2 Diabetes Screening Test by Means of a Pulse Oximeter

Monte

Lujan²,

Rusinol

et al. 2017

IEEE Trans. Biomed. Eng.

View full text Add to dashboard Cite

Abstract-In this paper, we propose a method for screening for the presence of type 2 diabetes by means of the signal obtained from a pulse oximeter. The screening system consists of two parts; the first analyses the signal obtained from the pulse oximeter, and the second consists of a machine-learning module.The system consists of a front end that extracts a set of features form the pulse oximeter signal. These features are based on physiological considerations. The set of features were the input of a machine-learning algorithm that determined the class of the input sample, i.e. whether the subject had diabetes or not. The machine-learning algorithms were random forests, gradient boosting, and linear discriminant analysis as benchmark. The system was tested on a database of 1, 157 subjects (two samples per subject) collected from five community health centres.The mean receiver operating characteristic (ROC) area found was 69.4% (median value 71.9% and range [75.4%−61.1%]), with a specificity=64% for a threshold that gave a sensitivity=65%.We present a screening method for detecting diabetes that has a performance comparable to the glycated haemoglobin (haemoglobin A1c HbA1c) test, does not require blood extraction, and yields results in less than five minutes.

show abstract

Section: ) Cepstral Analysismentioning

confidence: 99%

Type 2 Diabetes Screening Test by Means of a Pulse Oximeter

Monte

Lujan²,

Rusinol

et al. 2017

IEEE Trans. Biomed. Eng.

View full text Add to dashboard Cite

show abstract

“…By repeating the above EM iteration, we can get a series of approximate pdf whose mode is approaching to the mode 1 of the true posterior pdf (31) Thus, the hyperparameters are obtained at the last (actually th) EM iteration by using (13)- (19) to satisfy (32) and the CDHMM parameters are updated accordingly.…”

Section: M-stepmentioning

confidence: 99%

“…1, many existing techniques can be applied. They include, for example, the popular cepstral mean subtraction algorithm [2], different cepstral normalization methods (e.g., CDCN) discussed in [1], ML-based feature space stochastic matching methods [7], [41], [33], signal conditioning techniques [30], [31], etc. Acoustic normalization could even be integrated into the feature extraction stage, e.g., speaker normalization via vocal tract length normalization using frequency warping [39], [11], [22].…”

mentioning

confidence: 99%

On-line adaptive learning of the continuous density hidden Markov model based on approximate recursive Bayes estimate

Huo

Lee²

1997

IEEE Trans. Speech Audio Process.

View full text Add to dashboard Cite

“…In Appendix C, we show that the gradients of are given by (27) (28) (29) 1 If the number of classes, Y, is very large, one can restrict the sum over y in (22) to the N-best alternatives, where N Y.…”

Section: B Multivariate Gaussian Pdf'smentioning

confidence: 99%

“…In particular, for the th data point, let (22) denote the log-likelihood-ratio between the softmax-smoothed competing hypothesis 1 and the correct one. The function provides a smooth measure of the misclassification of .…”

Section: A Discriminative Trainingmentioning

confidence: 99%

Maximum likelihood and minimum classification error factor analysis for automatic speech recognition

Saul

Rahim

2000

IEEE Trans. Speech Audio Process.

Self Cite

View full text Add to dashboard Cite

Abstract-Hidden Markov models (HMM's) for automatic speech recognition rely on high-dimensional feature vectors to summarize the short-time properties of speech. Correlations between features can arise when the speech signal is nonstationary or corrupted by noise. We investigate how to model these correlations using factor analysis, a statistical method for dimensionality reduction. Factor analysis uses a small number of parameters to model the covariance structure of high dimensional data. These parameters can be chosen in two ways: 1) to maximize the likelihood of observed speech signals, or 2) to minimize the number of classification errors. We derive an expectation-maximization (EM) algorithm for maximum likelihood estimation and a gradient descent algorithm for improved class discrimination. Speech recognizers are evaluated on two tasks, one small-sized vocabulary (connected alpha-digits) and one medium-sized vocabulary (New Jersey town names). We find that modeling feature correlations by factor analysis leads to significantly increased likelihoods and word accuracies. Moreover, the rate of improvement with model size often exceeds that observed in conventional HMM's.

show abstract

Signal conditioning techniques for robust speech recognition

Cited by 49 publications

References 4 publications

Type 2 Diabetes Screening Test by Means of a Pulse Oximeter

Type 2 Diabetes Screening Test by Means of a Pulse Oximeter

On-line adaptive learning of the continuous density hidden Markov model based on approximate recursive Bayes estimate

Maximum likelihood and minimum classification error factor analysis for automatic speech recognition

Contact Info

Product

Resources

About