The Acoustic-Modeling Problem in Automatic Speech Recognition.

The Relative Distance Vector Neural Network (RDVNN) Model: A Hybrid Approach to Speech Recognition

A high performance neural-network-based speech recognition system is presented. A new approach towards feature representation for speech recognition, named state transition matrix (STM), is proposed to address temporal varying problem in speech recognition. Using STM, we need only a single-layer perceptron neural network to perform speech recognition. Experimental results show that an overall accuracy of 95% and 87% was achieved for speaker-dependent isolated word recognition and multi-speaker-dependent isolated word recognition, respectively.

In this paper, we present a new technique for social mobile robot speech recognition based on scene-associated training models. The key contribution of the paper is a real-time framework that reduces the effect of room reverberation and ambient noise, a challenging problem in speech recognition. In classical approaches, anechoic sound is used to train the model, with the main focus on removing reverberation or noise from the sound. Our technique differs in that we train a number of speech recognizers directly from the reverberated sound, by associating each recognizer with a unique visual scene, to deal with the varying reverberation properties of different rooms. By extracting local features from a captured image and recognizing a scene, the robot can use the appropriate speech recognizer that is trained for the particular structural properties of that scene. We tested our method by using a baseline speech recognition model (HTK) across a variety of rooms and different levels of background noise. The results show that the association between a visual scene and a corresponding speech recognizer greatly improves the robot's speech recognition accuracy, together with increasing the computational speed of recognition, compared to competing techniques.

In this paper, we describe the statistical approach to text translation. Starting from the Bayes decision rule as in speech recognition, we study the translation problem and present algorithms for tackling the translation task. The statistical translation approach uses two types of information: a translation model and a language model. The language model used is a bigram or trigram model as in speech recognition. The translation model is decomposed into a lexicon model and an alignment model. The alignment model is similar to the time alignment problem in speech recognition. After presenting the details of the alignment model, we describe the search problem and present suitable search algorithms, namely an iterative search method and a dynamic programming based method.

Human computer interaction (HCI) is very crucial in our day-to-day activity. Speech is one of the essential and intuitive ways to interact with machines such as Smartphone, which has multiple sensors as microphone, camera, etc. An efficient performance speech recognition system improves interaction between man and machines by making latter more receptive to user needs. Such system has Automatic speech recognition (ASR) engine, which is facing a unique challenge of accuracy in recognition rate. By integrating acoustic signal feature vectors with the visual features, a more robust audiovisual speech recognition engine (AVSR) could be developed for real environmental scenarios. This paper presents past research and development in the field of ASR and AVSR technologies. It describes key technological perspective and admiration of the fundamental progress in ASR and AVSR. The objective of this review is to summarize and compare some of the well-known methods experimented by previous researchers, and to conclude with direction on future research proficiency in HCI system using ASR and AVSR engine.

Continuous density hidden Markov models (CD-HMMs) are an essential component of modern systems for automatic speech recognition (ASR). These models assign probabilities to the sequences of acoustic feature vectors extracted by signal processing of speech waveforms. In this chapter, we investigate a new framework for parameter estimation in CD-HMMs. Our framework is inspired by recent parallel trends in the fields of ASR and machine learning. In ASR, significant improvements in performance have been obtained by discriminative training of acoustic models. In machine learning, significant improvements in performance have been obtained by discriminative training of large margin classifiers. Building on both these lines of work, we show how to train CD-HMMs by maximizing an appropriately defined margin between correct and incorrect decodings of speech waveforms. We start by defining an objective function over a transformed parameter space for CD-HMMs, then describe how it can be optimized efficiently by simple gradient-based methods. Within this framework, we obtain highly competitive results for phonetic recognition on the TIMIT speech corpus. We also compare our framework for large margin training to other popular frameworks for discriminative training of CD-HMMs.

Learning sets of filters using back-propagation

Introduction The ability to perceive and comprehend speech is, as we argued in the previous chapter, one of the human brain's more astonishing evolutionary accomplishments. Engineers and computer scientists have sought to emulate human speech recognition for about four decades, but it may be as many more before the best automatic speech recognition device can perform as well as the average five-year-old, though great strides have been made in recent years. Recognizing speech and the identity of the speaker is an ability that we normally take for granted, unless we are unfortunate enough to lose this vital skill temporarily or permanently as the result of ‘a stroke’ (cerebrovascular accident) or some other form of damage to language-critical areas of the brain. In this and the following two chapters we will be concerned with the early or peripheral stages of spoken language comprehension: with auditory signal processing; with the extraction of phonetic features that make up the ‘sound shapes’ of words; with how the phonological forms of words are retrieved from lexical memory and how these ‘sound traces’ of words may be represented in the recognition lexicon. We will leave to later chapters questions of how words are put together to form phrases or sentences, which belong to later stages of the spoken language comprehension process.

We show the effect of reverberation on the speech recognition performance in a far-field microphone. Given the reference of a close-talk microphone signal, an improvement is shown using an asymmetric non-causal inverse filter both in a synthetic and real room environment. Its variants in time and frequency domain are also presented and compared with other existing techniques. We argue for the approaches which specifically consider recognition performance as a goal in deriving the dereverberating schemes, with evaluation on real room recording recognition, as a future direction for solving reverberation problems in speech recognition.

The performance of current speech recognition systems is far below that of humans. Neural nets offer the potential of providing massive parallelism, adaptation, and new algorithmic approaches to problems in speech recognition. Initial studies have demonstrated that multilayer networks with time delays can provide excellent discrimination between small sets of pre-segmented difficult-to-discriminate words, consonants, and vowels. Performance for these small vocabularies has often exceeded that of more conventional approaches. Physiological front ends have provided improved recognition accuracy in noise and a cochlea filter-bank that could be used in these front ends has been implemented using micro-power analog VLSI techniques. Techniques have been developed to scale networks up in size to handle larger vocabularies, to reduce training time, and to train nets with recurrent connections. Multilayer perceptron classifiers are being integrated into conventional continuous-speech recognizers. Neural net architectures have been developed to perform the computations required by vector quantizers, static pattern classifiers, and the Viterbi decoding algorithm. Further work is necessary for large-vocabulary continuous-speech problems, to develop training algorithms that progressively build internal word models, and to develop compact VLSI neural net hardware.

Problems in speech recognition are often apparent in telecommunication situations. For ecologically valid assessments of such conditions, it is important to quantify the impact of real environments including acoustic conditions at a far-end communication device and all paths of transmission degradation. This study presents an automated matrix sentence test procedure based on automatic speech recognition (ASR) integrated in a Voice over Internet Protocol (VoIP) infrastructure and compares the individual effects of transmission degradations with results from laboratory measurements. Speech recognition thresholds (SRTs) were measured in 16 normal-hearing subjects in four test conditions: (a) a laboratory condition guided by a human experimenter, (b) a laboratory condition with reduced bandwidth and (c) additionally reduced headset quality to simulate typical communication systems, and (d) an automated, ASR-controlled adaptive test procedure over a real VoIP infrastructure. Errors of the ASR system were analyzed to show possible effects on measurement outcome Results: Measured SRTs showed a highly significant correlation (r = .93) between the fully automatic and "laboratory" conditions, with a constant bias of about 1 dB indicating a linear shift of the data without affecting the distribution around the mean. The individual impact of the different system degradations on SRTs could be quantified Conclusions: This study provides a proof of concept for automated ASR-based SRT measurements over VoIP systems for speech audiometric testing in real communication systems, as it produced results comparable to traditional laboratory settings for this group of 16 normal-hearing subjects. This makes VoIP services a promising candidate for speech audiometric testing in real communication systems.

Summary form only given, as follows. The possibility of automatic speech recognition on computers has fuelled dreams for many years. Indeed, automatic speech recognition holds great promise if the spoken words can be recognised correctly by machines in fluent speech. We review the current state-of-the-art in automatic speech recognition and point to the new directions that research in this field must explore to continue this progress. Statistical modelling of speech using utterances spoken by many speakers in a variety of environments has been important in achieving the progress that has been realised so far. But on the negative side, the statistical modelling approach has led to solutions that are narrow and do not generalise. The statistical approach, although fruitful in the early development of the technology, is now a hindrance as we become much more ambitious in seeking solutions to bigger problems. We take a fresh look at the problem of automatic speech recognition, not as a problem in statistical modelling, but as a problem in voice communication where the goal is to recognise every spoken word correctly and discuss the fundamental underpinnings of this new approach.

A model-based framework of classification error rate estimation is proposed for speech and speaker recognition. It aims at predicting the run-time performance of a hidden Markov model (HMM) based recognition system for a given task vocabulary and grammar without the need of running recognition experiments using a separate set of testing samples. This is highly desirable both in theory and in practice. However, the error rate expression in HMM-based speech recognition systems has no closed form solution due to the complexity of the multi-class comparison process and the need for dynamic time warping to handle various speech patterns. To alleviate the difficulty, we propose a one-dimensional model-based misclassification measure to evaluate the distance between a particular model of interest and a combination of many of its competing models. The error rate for a class characterized by the HMM is then the value of a smoothed zero-one error function given the misclassification measure. The overall error rate of the task vocabulary could then be computed as a function of all the available class error rates. The key here is to evaluate the misclassification measure in terms of the parameters of environmental-matched models without running recognition experiments, where the models are adapted by very limited data that could be just the testing utterance itself. In this paper, we show how the misclassification measure could be approximated by first computing the distance between two mixture Gaussian densities, then between two HMMs with mixture Gaussian state observation densities and finally between two sequences of HMMs. The misclassification measure is then converted into classification error rate. When comparing the error rate obtained in running actual experiments and that of the new framework, the proposed algorithm accurately estimates the classification error rate for many types of speech and speaker recognition problems. Based on the same framework, it is also demonstrated that the error rate of a recognition system in a noisy environment could also be predicted.

Ensuring the best quality and performance of modern speech technologies, today, is possible based on the widespread use of machine learning methods. The idea of this project is to study and implement an end-to-end system of automatic speech recognition using machine learning methods, as well as to develop new mathematical models and algorithms for solving the problem of automatic speech recognition for agglutinative (Turkic) languages. Many research papers have shown that deep learning methods make it easier to train automatic speech recognition systems that use an end-to-end approach. This method can also train an automatic speech recognition system directly, that is, without manual work with raw signals. Despite the good recognition quality, this model has some drawbacks. These disadvantages are based on the need for a large amount of data for training. This is a serious problem for low-data languages, especially Turkic languages such as Kazakh and Azerbaijani. To solve this problem, various methods are needed to apply. Some methods are used for end-to-end speech recognition of languages belonging to the group of languages of the same family (agglutinative languages). Method for low-resource languages is transfer learning, and for large resources – multi-task learning. To increase efficiency and quickly solve the problem associated with a limited resource, transfer learning was used for the end-to-end model. The transfer learning method helped to fit a model trained on the Kazakh dataset to the Azerbaijani dataset. Thereby, two language corpora were trained simultaneously. Conducted experiments with two corpora show that transfer learning can reduce the symbol error rate, phoneme error rate (PER), by 14.23 % compared to baseline models (DNN+HMM, WaveNet, and CNC+LM). Therefore, the realized model with the transfer method can be used to recognize other low-resource languages.

Current Problems in Automatic Speech Recognition