Automatic diacritization of Arabic for acoustic modeling in speech recognition

This paper describes the combination of two recent machine learning techniques for acoustic modeling in speech recognition: deep neural networks (DNNs) and graph-based semi-supervised learning (SSL). While DNNs have been shown to be powerful supervised classifiers and have achieved considerable success in speech recognition, graph-based SSL can exploit valuable complementary information derived from the manifold structure of the unlabeled test data. Previous work on graph-based SSL in acoustic modeling has been limited to frame-level classification tasks and has not been compared to, or integrated with, state-of-the-art DNN/HMM recognizers. This paper represents the first integration of graph-based SSL with DNN based speech recognition and analyzes its effect on word recognition performance. The approach is evaluated on two small vocabulary speech recognition tasks and shows a significant improvement in HMM state classification accuracy as well as a consistent reduction in word error rate over a state-of-the-art DNN/HMM baseline.

This paper describes a data-driven technique for optimizing the acoustic models for speech recognition systems that target commercial applications over telephones. Frame-averaged foreground log-likelihoods (foreground scores) correlate to recognition errors. These scores are used together with gender to optimize data weighting for the acoustic model. This process is interpreted as increasing the priors and associated parameters for poorly modeled data. The score-based optimization leads to about 7% fewer semantic errors on a live evaluation set collected after the last data used to estimate the acoustic model.

We study large-scale kernel methods for acoustic modeling in speech recognition and compare their performance to deep neural networks (DNNs). We perform experiments on four speech recognition datasets, including the TIMIT and Broadcast News benchmark tasks, and compare these two types of models on frame-level performance metrics (accuracy, cross-entropy), as well as on recognition metrics (word/character error rate). In order to scale kernel methods to these large datasets, we use the random Fourier feature method of Rahimi and Recht [2007]. We propose two novel techniques for improving the performance of kernel acoustic models. First, in order to reduce the number of random features required by kernel models, we propose a simple but effective method for feature selection. The method is able to explore a large number of non-linear features while maintaining a compact model more efficiently than existing approaches. Second, we present a number of frame-level metrics which correlate very strongly with recognition performance when computed on the heldout set; we take advantage of these correlations by monitoring these metrics during training in order to decide when to stop learning. This technique can noticeably improve the recognition performance of both DNN and kernel models, while narrowing the gap between them. Additionally, we show that the linear bottleneck method of Sainath et al. [2013a] improves the performance of our kernel models significantly, in addition to speeding up training and making the models more compact. Together, these three methods dramatically improve the performance of kernel acoustic models, making their performance comparable to DNNs on the tasks we explored.

Modeling heterogeneous data sources remains a fundamental challenge of acoustic modeling in speech recognition. We call this the multi-condition problem because the speech data come from many different conditions. In this paper, we introduce the fundamental confusability problem in multi-condition learning, then discuss the problem formalization, the taxonomy, and the architectures for multi-condition learning. While the ideas presented are applicable to all classifiers, we focus our attention in this work on acoustic models based on deep neural networks (DNN). We propose four different strategies for multi-condition learning of a DNN that we refer to as a mixed-condition model, a condition-dependent model, a condition-normalizing model, and a condition-aware model. Based on the experimental results on the voice search and short message dictation task and the Aurora 4 task, we show that the confusability introduced when modeling heterogeneous data depends on the source of acoustic distortion itself, the front-end feature extractor, and the classifier. We also demonstrate the best approach for dealing with heterogeneous data may not be to let the model sort it out blindly, even with a classifier as sophisticated as a DNN.

We present in this chapter a practical approach in building Arabic automatic speech recognition (ASR) system for mobile telecommunication service applications. We also present a procedure in conducting acoustic modelling adaptation to better take into account the pronunciation variation across the Arabic speaking countries. Modern Standard Arabic (MSA) is the common spoken and written language for all the Arab countries, ranging from Morocco in the west to Syria in the East, including Egypt, and Tunisia. However, the pronunciation varies significantly from one country to another to a degree that two persons from different countries may not be able understand each other. This is because Arabic speaking countries are characterized by a large number of dialects that differ to an extent that they are no longer mutually intelligible and could almost be described as different languages. Arabic dialects are often spoken rather than written varieties. MSA is common across the Arab countries, but it is often influenced by the dialect of the speaker. This particularity of the Arabic countries constitutes a practical problem in the development of a speech-based application in this region; suppose a speech application system is built for one country influenced by one dialect, what does it take to adapt the system to serve another country with a different dialect region? This is particularly challenging since resource to build accurate speaker independent Arabic ASR system for mobile telecommunication service applications are limited for most of the Arabic dialects and countries. Recent advances in speaker independent automatic speech recognition (SI-ASR) have demonstrated that highly accurate recognition can be achieved, if enough training data is available. However, the amount of available speech data that take into account the dialectal variation of each Arabic country is limited, making it challenging to build a high performance SI-ASR system, especially when we target specific applications. Another big challenge when building an SI-ASR is to handle speaker variations in spoken language. These variations can be due to age, gender, educational level as well as the dialectical variants of Arabic language. Usually an ASR system trained in one regional variation exhibits poorer performance when applied to another regional variation. Three problems may arise when a SI-ASR system built for one dialect but applied to target users with a different dialect: (1) Acoustic model mismatch, (2) Pronunciation lexicon mismatch and (3) Language model mismatch.

The focus of this dissertation is to derive and demonstrate effective stochastic models for the speech recognition problem. Acoustic modeling for speech recognition typically involves representing the speech process within stochastic models. Modeling this high frequency time series effectively is a fundamental problem. This dissertation devised an objective function that relates the true speech distribution to its estimate. It is shown that through optimizing this function the speech process time series can be modeled without loss of information. The thesis proposes two such models that are developed to optimize the devised objective function. The first an acoustic model formulated for the speech with noise problem. The second a discriminately trained model consisting of optimal discriminant ML estimators. The first, a combination of recognizers that through a simple system fusion, combines multiple speech processes at the decision level. This is a stochastic modeling method devised to combine a parameterized spectral missing data, MD, theory based and a cepstral based speech process using a coupled hidden variable topology. In using a fused coupled hidden Markov model, HMM, topology, an optimal acoustic model is proposed that is inherently more robust than single process models under noisy conditions. The theoretical capability of this model is tested under both stationary and non stationary noise conditions. Under these test conditions the fused model has greater recognition accuracies than those of single process models. The second, formulated with a methodology that segments the acoustic space appropriately for discriminately trained models that optimize the devised objective function. This acoustic space is modeled with discriminant ML estimators formed with optimal decision boundaries using the large margin, support vector machine, SVM, learning method. These discriminately trained models maximize the entropy of the observation space and thereby are capable to model the speech process without loss. This is demonstrated experimentally with frame level classification error rates that are ∼ ≤ 3%.

Despite their known weaknesses, hidden Markov models (HMMs) have been the dominant technique for acoustic modeling in speech recognition for over two decades. Still, the advances in the HMM framework have not solved its key problems: it discards information about time dependencies and is prone to overgeneralization. In this paper, we attempt to overcome these problems by relying on straightforward template matching. The basis for the recognizer is the well-known DTW algorithm. However, classical DTW continuous speech recognition results in an explosion of the search space. The traditional top-down search is therefore complemented with a data-driven selection of candidates for DTW alignment. We also extend the DTW framework with a flexible subword unit mechanism and a class sensitive distance measure-two components suggested by state-of-the-art HMM systems. The added flexibility of the unit selection in the template-based framework leads to new approaches to speaker and environment adaptation. The template matching system reaches a performance somewhat worse than the best published HMM results for the Resource Management benchmark, but thanks to complementarity of errors between the HMM and DTW systems, the combination of both leads to a decrease in word error rate with 17% compared to the HMM results

Nonparametric Bayesian models have become increasingly popular in speech recognition for their ability to discover data’s underlying structure in an iterative manner. Dirichlet process mixtures (DPMs) are a widely used nonparametric method that do not require a priori assumptions about the structure of the data. DPMs, however, require an infinite number of parameters so inference algorithms are needed to make posterior calculations tractable. The focus of this work is an evaluation of three variational inference algorithms for acoustic modeling: Accelerated Variational Dirichlet Process Mixtures (AVDPM), Collapsed Variational Stick Breaking (CVSB), and Collapsed Dirichlet Priors (CDP).

Training very deep neural networks is very difficult because of gradient degradation. However, the incomparable expressiveness of the many deep layers is highly desirable at testing time and usually leads to better performance. Recently, training techniques such as residual networks that enable us to train very deep networks have proved to be a great success. In this paper, we studied the application of the recently proposed deep networks with stochastic depth (DNSD) to train deeper acoustic models for speech recognition. By randomly dropping a subset of layers during training, the studied stochastic depth training method helps reduce the training time substantially, yet the networks trained are much deeper since all the layers are kept during testing. We investigated this approach on the TIMIT data set. Our preliminary experimental results show that when training data are limited, deep networks with stochastic depth helps very little. However, when more training data are available, DNSD significantly improves the recognition accuracy, compared with a conventional deep neural networks.

Acoustic modeling in large vocabulary continuous speech recognition systems is commonly done by building the models for subword units such as phonemes, syllables or senones. In recent years, various end-to-end systems using acoustic models built at grapheme or phoneme level have also been explored. These systems either require a lot of data and/or heavily rely on the use of language models or pronunciation dictionary for good recognition performance. With the intention of reducing the dependence on data or external models, we have explored the usage of phonetic features in building acoustic models for speech recognition. The phonetic features describe a sound based on the speech production mechanism in humans. Multi-label classification models are built for detection of phonetic features in a given speech signal. The detected phonetic features are used along with the acoustic features as input to models for phoneme identification. The effectiveness of the proposed approach is demonstrated on TIMIT and Wall Street Journal corpora. Performance improvement over other phoneme recognition studies using the phonetic features is obtained.

Maxout neurons for deep convolutional and LSTM neural networks in speech recognition

Although the Arab world has an estimated number of 250 million Arabic speakers, there has been little research on Arabic speech recognition when compared to other languages of similar importance (e.g. Mandarin). Due to the lack of diacritic Arabic text and the lack of Pronunciation Dictionary (PD), most of previous work on Arabic Automatic Speech Recognition has been concentrated on developing recognizers using Romanized characters i.e. let the system recognizes the Arabic word as an English one, then map it to Arabic word from lookup table that maps the Arabic word to its Romanized pronunciation.

Automatic speech recognition is a research field for enabling computers to receive voice feedback from humans and to translate it with the greatest probability of correctness. Speech is the easiest way to communicate the message in basic context. Speech recognition is referring the allowing of machines to perceive and accept speech information from humans for the greatest probability of accuracy. Speech recognition will play an ever more important role in the future of human with machine interfaces. Arabic speech recognition (ASR) is a most complex job and has received little attention than other languages. In this paper we reviewing the most important deep learning methods used in speech recognition in Arabic language. Through the search the main finding in this review that the Arabic is one of the most spoken languages and least highlighted in terms of speech recognition. One of the emerging techniques is using neural networks with deep learning for speech recognition. Deep Neural Networks (DNNs) have reached to suitable performance, Recurrent Neural Networks (RNN) is one of the best models applied for sequential data and convolution neural network (CNN) has obtained a best accuracy rate in speech recognition.

This work focuses on generating children's HMM-based acoustic models for speech recognition from adult acoustic models. Collecting children's speech data is more costly compared to adult's speech. The patent-pending method developed in this work requires only adult data to estimate synthetic children's acoustic models in any language and works as follows: For a new language where only adult data is available, an adult male and an adult female model is trained. A linear transformation from each male HMM mean vector to its closest female mean vector is estimated. This transform is then scaled to a certain power and applied to the female model to obtain a synthetic children's model. In a pronunciation verification task the method yields 19% and 3.7% relative improvement on native English and Spanish children's data, respectively, compared to the best adult model. For Spanish data, the new model outperforms the available real children's data based model by 13% relative.

As one of the core modules of English speech recognition system, the performance of acoustic model directly affects the recognition effect of the final system. Aiming at the problems of acoustic modeling, model optimization and training efficiency in English speech recognition system, this paper focuses on the acoustic modeling method based on deep learning technology. By analyzing the basic theory of deep neural network and HMM, the structure and parameter configuration of DNN-HMM are intensively studied. A new hybrid network model is proposed by using clustered state instead of single factor state as the output unit of the neural network, and it is applied to the acoustic model of English speech recognition. The simulation results show that the improved scheme can get better recognition effect by increasing the enhancement of speech features under the condition of three-phoneme structure. In addition, the experimental results also verify that the acoustic modeling method based on DNN-HMM is superior to traditional GMM-HMM method.