Automatic speech signal segmentation based on the innovation adaptive filter

Abstract

Speech segmentation is an essential stage in designing automatic speech recognition systems and one can find several algorithms proposed in the literature. It is a difficult problem, as speech is immensely variable. The aim of the authors' studies was to design an algorithm that could be employed at the stage of automatic speech recognition. This would make it possible to avoid some problems related to speech signal parametrization. Posing the problem in such a way requires the algorithm to be capable of working in real time. The only such algorithm was proposed by Tyagi et al., (2006), and it is a modified version of Brandt's algorithm. The article presents a new algorithm for unsupervised automatic speech signal segmentation. It performs segmentation without access to information about the phonetic content of the utterances, relying exclusively on second-order statistics of a speech signal. The starting point for the proposed method is time-varying Schur coefficients of an innovation adaptive filter. The Schur algorithm is known to be fast, precise, stable and capable of rapidly tracking changes in second order signal statistics. A transfer from one phoneme to another in the speech signal always indicates a change in signal statistics caused by vocal track changes. In order to allow for the properties of human hearing, detection of inter-phoneme boundaries is performed based on statistics defined on the mel spectrum determined from the reflection coefficients. The paper presents the structure of the algorithm, defines its properties, lists parameter values, describes detection efficiency results, and compares them with those for another algorithm. The obtained segmentation results, are satisfactory.

Hybrid methods of brandt’s generalised likelihood ratio and short-term energy for malay word speech segmentation

“…Therefore 0.4 second framing is relatively effective for this type of segmentation. Pantun five (5) shows the worst accuracy as a lot of the words are made of prefixes such as dihati and membujang. Insertion occurs in between the prefix and the word hence lowering the accuracy of it.…”

“…The optimal boundary is assumed to be the frame that has maximum likelihood of the GMM model associated with the CART leaf node for the pseudo-triphone. Experimental results in [5] shows that the refined HMM is more accurate than the standard HMM segmentation.…”

“…Speech segmentation can be defined as the process of finding the limits (with specific characteristic) in natural spoken language between words, syllables or phonemes. [4,5]. The main objective of Speech segmentation is to serve other speech analysis problems such as speech synthesis, data training for speech recognizers, or to fabricate and label prosodic databases.…”

“…[10] In Automatic Speech Recognition ASR systems, segmentation can be performed: (i) At the system training stage, when segmentation is applied to the training set recordings. (ii) At the recognition stage [5].…”

A Bayesian binary algorithm for root mean squared-based acoustic signal segmentation

“…The problem of signal segmentation arises in different contexts [19,28,22,16,27]. The problem is broadly defined as follows: given a discretely sampled signal y ∈ N , divide it in contiguous sections that are internally homogeneous with respect to some characteristic.…”