Large-vocabulary Speaker-independent Continuous Speech Research Articles

Synopsis on Arabic speech recognition

With the advancement and increased usage of intelligible smart devices, researchers have an intensified interest in the field of large-vocabulary speaker-independent continuous speech recognition. Although considerable research has been devoted to English speech recognition, less attention has been paid to the Arabic speech recognition. This paper aims to highlight the achievements that have been made during the last several decades of Arabic speech recognition. The paper also discusses speech recognition components such as corpora, phonemes, language models, acoustic models, and performance evaluation. For an empirical evaluation of Arabic speech recognition, the free, off-the-shelf Mac Soundflower tool was employed to evaluate the recognition performance using a continuous speech corpus that contains 2.63 h (by 10 male and 10 female speakers) of modern standard Arabic (MSA) broadcast news. The experimental results indicate recognition accuracy at 54.02%, and the accuracies for the male and female speakers are almost the same. This result promotes the need for further research to expose the practical range of accuracy. The performance’s decline might be an indication of the necessity for further research to boost the overall recognition accuracy.

Spoken language recognition on a DSP array processor

A new architecture is presented to support the general class of real-time large-vocabulary speaker-independent continuous speech recognizers incorporating language models. Many such recognizers require multiple high-performance central processing units (CPU's) as well as high interprocessor communication bandwidth. This array processor provides a peak CPU performance of 2.56 giga-floating point operations per second (GFLOPS) as well as a high-speed communication network. In order to efficiently utilize these resources, algorithms were devised for partitioning speech models for mapping into the array processor. Also, a novel scheme is presented for a functional partitioning of the speech recognizer computations. The recognizer is functionally partitioned into six stages, namely, the linear predictive coding (LPC) based feature extractor, mixture probability computer, (phone) state probability computer, word probability computer, phrase probability computer, and traceback computer. Each of these stages is further subdivided as many times as necessary to fit the individual processing elements (PE's). The functional stages are pipelined and synchronized with the frame rate of the incoming speech signal. This partitioning also allows a multistage stack decoder to be implemented for reduction of computation. >

A comparative study of discrete, semicontinuous, and continuous hidden Markov models

In this paper, we first extended the semicontinuous hidden Markov model to incorporate multiple code-books. The robustness of the semicontinuous output probability is enhanced by the combination of multiple codewords and multiple codebooks. In addition, we compared the semicontinuous model with the continuous mixture model and the discrete model in a large-vocabulary speaker-independent continuous speech recognition (DARPA resource management) task. The model assumption and parameter size issues are addressed in particular through these experiments. When the acoustic parameters are not well modelled by the continuous probability density, the model assumption problems may cause the recognition accuracy of the semicontinuous model or the continuous mixture model to be inferior to the discrete model. We also found that the SCHMM can have a large number of free parameters in comparison with the discrete HMM because of its smoothing ability. With explicit male and female clustered models and for conditional feature sets, we were able to reduce the error rate of discrete-model-based SPHINX by more than 20%.

Recent progress and future outlook of the SPHINX speech recognition system

This paper describes recent and future work on the sphinx speech recognition system at Carnegie Mellon. Sphinx is a large-vocabulary speaker-independent continuous speech recognizer based on discrete hidden Markov models. Sphinx uses multiple codebooks of LPC-derived features, and phonetic units that model function words and phrases, as well as within and between word generalized triphones. Sphinx is first trained with the Forward-Backward algorithm, and then with the corrective training algorithm. On a 997-word task, sphinx attained accuracies of 96% with a grammar (perplexity 60), and 82% without grammar (perplexity 997). This paper describes the sphinx System and its performance in detail, as well as outlines our future work.

On large-vocabulary speaker-independent continuous speech recognition

In this paper, we describe sphinx, the world's first accurate large-vocabulary speaker-independent continuous speech recognition system. We will present current results of sphinx, compare its performance against similar systems, and account for its high accuracy.