SVMs for Automatic Speech Recognition: A Survey

Abstract

Hidden Markov Models (HMMs) are, undoubtedly, the most employed core technique for Automatic Speech Recognition (ASR). Nevertheless, we are still far from achieving high-performance ASR systems. Some alternative approaches, most of them based on Artificial Neural Networks (ANNs), were proposed during the late eighties and early nineties. Some of them tackled the ASR problem using predictive ANNs, while others proposed hybrid HMM/ANN systems. However, despite some achievements, nowadays, the preponderance of Markov Models is a fact.During the last decade, however, a new tool appeared in the field of machine learning that has proved to be able to cope with hard classification problems in several fields of application: the Support Vector Machines (SVMs). The SVMs are effective discriminative classifiers with several outstanding characteristics, namely: their solution is that with maximum margin; they are capable to deal with samples of a very higher dimensionality; and their convergence to the minimum of the associated cost function is guaranteed.These characteristics have made SVMs very popular and successful. In this chapter we discuss their strengths and weakness in the ASR context and make a review of the current state-of-the-art techniques. We organize the contributions in two parts: isolated-word recognition and continuous speech recognition. Within the first part we review several techniques to produce the fixed-dimension vectors needed for original SVMs. Afterwards we explore more sophisticated techniques based on the use of kernels capable to deal with sequences of different length. Among them is the DTAK kernel, simple and effective, which rescues an old technique of speech recognition: Dynamic Time Warping (DTW). Within the second part, we describe some recent approaches to tackle more complex tasks like connected digit recognition or continuous speech recognition using SVMs. Finally we draw some conclusions and outline several ongoing lines of research.

Automatic Recognition of Speech-evoked Brainstem Responses to English Vowels

“…However, since this was a first attempt, we only investigated a limited number of feature extraction and pattern recognition methods. There are other methods available either in terms of feature extraction such as relative spectral filtering of log domain coefficients (Shrawankar and Thakare, 2010), or in terms of pattern recognition methods such as support vector machines (Solera-Urena et al, 2007) or hybrid models of HMM and ANN (Trentin and Gori, 2001) that can be examined to identify the best possible method for achieving a higher recognition rate for speech-evoked ABR signals. Also since speech-evoked ABRs are typically the responses to the single stimuli (there is no transition of states between stimuli), it could be more beneficial to use the Gaussian Mixture Model (GMM) that is a parametric probability density function, characterized as a weighted sum of Gaussian component densities (Vyas, 2013), instead of HMM that models the data as a sequence of states.…”

Reconocimiento de voz basado en MFCC, SBC y Espectrogramas

“…Este algoritmo fue propuesto en 1992 y es una versión no lineal de un algoritmo lineal mucho más antiguo, la regla de decisión del hiperplano óptimo, que fue introducido en los años sesenta [6]. En la ecuación 10 se muestra la fórmula general de la SVM y en [6] se encuentran los kernels más usados.…”

“…En la ecuación 10 se muestra la fórmula general de la SVM y en [6] se encuentran los kernels más usados.…”

Real-Time Robust Automatic Speech Recognition Using Compact Support Vector Machines

“…For a more detailed review on the use of support vector machines for ASR, including systems and difficulties, refer to [33], [34].…”