Advances In Speech Processing Research Articles

Review of Advances in Speech Processing with Focus on Artificial Neural Networks

Speech is the primary way via which most humans communicate. Computers facilitate this transfer of information, especially when people interact with databases. While some methods to manipulate and interpret speech date back many decades (e.g., Fourier analysis), other processing techniques were developed late last century (e.g., linear predictive coding and hidden Markov models). Nonetheless, the last 25 years have seen major advances leading to the wide acceptance of computer-based speech processing, e.g., cellular telephones and real-time online conversations. This paper reviews older techniques and recent methods that focus largely on artificial neural networks. The major highlights in speech research are examined, without delving into mathematical detail, while giving insight into the research choices that have been made. The focus of this work is to understand how and why the discussed methods function well.

Hearing-impaired Children and Adults in Need of Cochlear Implant- A Hospital Based Cross-sectional Study

Introduction: Hearing is essential to learn language and speech and to develop cognitive skills. According to World Health Organisation (WHO), world-wide approximately 350 million people have hearing disorder. The performance of some moderately severe hearing-impaired adults and children using hearing aids is poorer than that of even profoundly hearing-impaired individuals using cochlear implants with advanced speech processing strategies. Aim: To find out the appropriate candidates in need of cochlear implantation in Prayagraj district of Uttar Pradesh, India. Materials and Methods: This cross-sectional study was conducted on hearing impaired patients attending Ear Nose Throat (ENT) Out Patient Department (OPD) of tertiary health care center in Prayagraj Uttar Pradesh from August 2011 to July 2012. The study population consisted of 95 patients categorised into four groups (pre-lingual children and adults, post-lingual children and adults) who presented with the chief complaints of impaired hearing and delayed speech. A detailed history, clinical and other relevant systemic examination and investigations were done. Patients were subjected to free field audiometry, pure tone audiometry and Brainstem-Evoked Response Audiometry (BERA) for assessment of hearing threshold level. Patients used hearing aid for six months and after hearing aid use, their hearing threshold and speech discrimination scores were estimated. Patients, whose aided score on open-set sentence test was less than 50%, were selected as candidate for cochlear implants, as they were not significantly benefited by hearing aids. Results: Out of the 95 cases, 48 were found to be suitable for cochlear implantation, which formed 50.5% of the study group. Pre-lingual hearing impaired adults had the highest percentage i.e., 86.7%, fulfilling the criteria for cochlear implantation followed by pre-lingual children, in whom 46.7% fell in this category. A 26.7% of postlingual children and 20% of postlingual adults were found suitable for cochlear implantation. Conclusion: The pre-lingual hearing impaired children and adults are more in need for cochlear implantation than the postlingual hearing-impaired children and adults. Approximately, 134,501 cases were estimated to be, in need of cochlear implantation in Prayagraj district based on the census estimate of 2011.

Capturing talk and proximity in the classroom: Advances in measuring features of young children's friendships

Young children's friendships fuel essential developmental outcomes (e.g., social-emotional competence) and are thought to provide even greater benefits to children with or at-risk for disabilities. Teacher and parent report and sociometric measures are commonly used to measure friendships, and ecobehavioral assessment has been used to capture its features on a momentary basis. In this proof-of-concept study, we use Ubisense, the Language ENvironmental Analysis (LENA) recorder, and advanced speech processing algorithms to capture features of friendship – child-peer speech and proximity within activity areas. We collected 12,332 1-second speech and location data points. Our preliminary results indicate the focal child at-risk for a disability and each playmate spent time vocalizing near one another across 4 activity areas. Additionally, compared to the Blocks activity area, the children had significantly lower odds of talking while in proximity during Manipulatives and Science. This suggests that the activity areas children occupy may affect their engagement with peers and, in turn, the friendships they development. The proposed approach is a groundbreaking advance to understanding and supporting children's friendships.

Feature-driven machine learning to improve early diagnosis of Parkinson's disease

Although advances in speech processing have facilitated the prognostic assessment of patients with Parkinson's Disease (PD), there is no objective method towards its early detection in a clinical setting. This study investigated the application of a novel hybrid Artificial Intelligence-based classifier for aiding early diagnosis of PD. Data on dysphonic measures and clinical scores on 68 subjects were obtained from the University of California-Irvine (UCI) Machine Learning database. Weights derived from a Multi-Layer Perceptron (MLP) were applied for feature selection and their moduli used to rank the input features based on their relative importance towards discriminating between physiological and pathological data patterns. Thus, the initial 27 features were reduced to 20 selected diagnostic factors. This reduced feature set was then input to a Lagrangian Support Vector Machine (LSVM) for classification. The overall performance of this hybrid feature-driven algorithm (MLP-LSVM) was thus compared against commercially available software and classifiers from similar studies. Results indicate an overall classification accuracy and the area under the receiver operating characteristic curve of 100% for the proposed feature-driven algorithm (MLP-LSVM), with relatively faster convergence, thus demonstrating its potential for aiding early diagnosis of PD in a clinical setting.

Unlocking the Potential of the Spoken Word

Advances in speech processing may soon place speech and writing on a more equal footing, with broad implications for many aspects of society.

Catering for disabled surfers – A case study in web site navigation for disabled students

INTRODUCTION Since the computer as we know it was first developed, man has dreamt of systems that can be operated by voice command. Science fiction has portrayed a future where users communicate with computers by voice, and generations have grown up expecting a new dawn whereby we speak and the computer speaks back. The days of communicating naturally with computers by speech, however, are not so far off. Recent technological developments have led to a new way of thinking about how we should interact with computers in general, with the development of the XML specification and the use of XML to develop applications for many types of data. In particular, Voice XML and Speech Application Language Tags (SALT) (http://www.saltforum.org/) are technologies that could revolutionise navigation of computer systems. Speech technology seems an ideal complement to web appliances. Advances in speech processing have already yielded for users a far more natural sounding voice reading web information. The incorporation of speech recognition can be an important source of user acceptance of any system. Teaching through the use of web pages has provided an alternative way of learning for many people including those who have a disability. A severely physically disabled student may find it advantageous to study at home, but if the principal teaching medium is the web and the student is unable to independently navigate the pages of the site, then a barrier to learning exists. The provision of services such as audio transcription may depend on the number of students who require the service, and providing 100 students per year with an audio recording of the same course does have advantage in scale compared to a situation where there are only a few students in a course. This paper reviews the technologies currently available for speech interaction with computers and suggests how the future of web navigation may benefit from these technologies. The paper also discusses the results obtained from evaluation of a voice navigatable website aimed at disabled users. DISABILITY Speech recognition is being used to give some severely disabled people a means of remote control of their environments--a way to choose independently among several activities such as watching television and listening to the radio. For people with physical disabilities affecting the use of their hands, productivity is far more personal. Speech recognition can enable some people to perform daily living tasks without assistance. For others, such as the growing number of professionals afflicted with Repetitive Stress Syndrome (RSI), speech recognition represents a means of getting or keeping employment. Reasons that people with a disability use voice recognition include limited use of hands, difficulty with spelling, inability to touch type, and need of a fast way of working. Therefore a user needing to work hands-free will need a reliable method of controlling menus: a means of moving and clicking the mouse using voice commands and a correction mechanism that works completely by voice. Users with spelling difficulties need predictive correction, a text to speech feature to help with proofreading, and a facility to play what was dictated (i.e. their voice). People who wish to speed up typing are likely to be most interested in speed of recognition and easy correction using the mouse. Many visually impaired users are attracted to the idea of operating their computer by voice, necessitated either by keyboarding difficulties or lack of touch-typing proficiency. Voice in/Voice out is the use of a computer with voice recognition as a means of information input, combined with synthetic speech that enables the user to know what is on the computer screen. Many problems are inherent in using voice recognition with speech output. The main problems are: * Hearing the words or phrases echoed back is often not enough for the user to be sure there are no errors in the recognition or formatting of the text. …

Cochlear implants in children: safety as well as speech and language

The development of cochlear implants for children at the University of Melbourne and the Bionic Ear Institute, has consisted of a routine of biological and engineering safety followed by evaluation of speech processing strategies on adults before they are undertaken on children. The initial safety studies were to ensure that insertion was atraumatic, the electrical stimulus parameters did not lead to loss of ganglion cells and that the electrode could be inserted without the risk of middle ear infection leading to meningitis. The initial second formant extraction scheme was shown to produce significant open-set speech understanding in adults and was approved by the US Food and Drug Administration (FDA) in 1985. Following this, an international study was undertaken for the FDA on children using a strategy that also included the first formant, and was approved in 1990. Additional advances in speech processing have been evaluated on adults. However, before using one with high rates of stimulation, it was tested for safety on experimental animals. Further advances have been anticipated in particular through the development of a peri-modiolar array, the Nucleus Contour. Prior to its use on adults, it was tested in the human temporal bone and found to lead to minimal trauma. It was evaluated in adults and found to lead to better current localization and lower thresholds. A study was undertaken in children using a spectral maxima scheme at high rates (advanced combination encoder (ACE)) and the Contour array as it had given best results in adults. It was approved as safe and effective for use in children in 2000. Studies were also undertaken to look at plasticity and visual dominance particularly through cognitive studies and the use of the McGurk effect. This demonstrated that deaf children with implants rely heavily on visual information and there is a great need to have unambiguous auditory stimuli to get best results.

Cochlear implants in Children: safety as well as speech and language

The development of cochlear implants for children at the University of Melbourne/Bionic Ear Institute has consisted of a routine of biological and engineering safety followed by evaluation of speech processing strategies on adults before they are undertaken on children. The initial safety studies were to ensure that insertion was atraumatic, the electrical stimulus parameters did not lead to loss of ganglion cells and that the electrode could be inserted without the risk of middle ear infection leading to meningitis.The initial second formant extraction scheme was shown to produce significant open set speech understanding in adults and was approved by the U.S. Food and Drug Administration (FDA) in 1985. Following this, an international study was undertaken for the FDA on children using a strategy that also included the first formant, and was approved in 1990.Additional advances in speech processing have been evaluated on adults. However, before using one with high rates os stimulation it was tested for safety on experimental animals. Further advances have been anticipated in particular through the development of a perimodiolar array, the Nucleus “Contour.” Prior to its use on adults, it was tested in the human temporal bone and found to lead to minimal trauma. It was evaluated in adults and found to lead to better current localization and lower thresholds. A study was undertaken in children using a spectral maxima scheme at high rates (ACE) and the contour array as it had given best results in adults. It was approved as safe and effective for use in children in 2000.Studies have also been undertaken to look at plasticity and visual dominance particularly through cognitive studies and the use of the McGurk effect. This demonstrated that deaf children with implants rely heavily on visual information and there is a great need to have unambiguous auditory stimuli to get best results.

A source–filter model of humpback whale (Megaptera novaeangliae) vocal sound production

Acoustic waveforms produced by vocalizing humpback whales are well described by a source–filter model. This model characterizes vocalizations as the output of a time-varying signal generator (the source) passed through a time-varying resonator (the filter). For most humpback whale vocalizations, the source can be modeled as either a quasiperiodic impulse train, white noise, or a combination of these two signal types; the resonator can be modeled as an all-pole filter. The source–filter model is useful because it (a) provides a way to quantitatively characterize humpback whale vocalizations that can be automated and that allows for the ‘‘resynthesis’’ of sounds; (b) facilitates the application of advanced speech processing technologies to the analysis of humpback vocalizations; and (c) can potentially provide clues about the actual mechanisms humpback whales use to produce sounds. Currently, it is not known how humpback whales internally produce sounds, although structures in the larynx are thought to be involved. Source–filter-based analysis of humpback vocalizations revealed characteristic regions of frequency enhancement (formants) that may correspond to resonant frequencies of a physiological ‘‘filter.’’ Distribution of formants varied with pitch in a manner that appeared similar to pitch-related formant variation in human singers, suggesting that both species use analogous production mechanisms.

The CallManager system: A platform for intelligent telecommunications services

Bellcore's CallManager system is an experimental prototype of a call screening and `follow-me anywhere' service platform that makes use of a personal appointment calendar and a personal phonebook (accessed through the customer's wireline or wireless PDA) to determine the customer's current location and the importance of each caller to the customer. This is an example of a future class of network-provided services that utilize advanced speech processing and customer specific data to greatly enhance the value of the telephone network to the customer. Such services allow important calls to reach the customer while screening out unwanted or unimportant calls. These services can potentially also proactively provide information to a customer `as it happens' wherever the customer may happen to be.

Stability analysis on the time-domain finite-difference solution of a basilar-membrane vibration model: Application to acoustic signal processing

In the past several years, models for nonlinear wave motions in the basilar membrane based on the hydrodynamic principle and on the biophysical mechanisms of the cochlear function have been developed. The significance of such models can be projected to be their potential applications in designing front-end components of advanced speech processing systems, as well as in serving a powerful tool for studying information processing in central parts of the auditory system. Unfortunately, computer simulations of the models in the past had been exceedingly slow and sometimes unstable, which had prevented use of the models in many possible speech-processing applications. In this paper numerical properties of a finite-difference scheme for the time-domain solution of a one-dimensional nonlinear transmission line basilar membrane model are studied. In particular, the von Neumann method is applied to obtain a sufficient condition under which the finite-difference solution to the model is guaranteed to be stable. The utility of this condition is that it determines the optimal time and spatial mesh sizes, thereby guaranteeing a minimal amount of computation in obtaining stable and accurate model outputs. Also presented are preliminary results on applying the basilar-membrane model, based on the most efficient computation schemes derived from the stability analysis, to process speech signals.

Opportunities for advanced speech processing in military computer-based systems

The author presents a study of military applications of advanced speech processing technology which includes three major elements: (1) review and assessment of current efforts in military applications of speech technology. (2) identification of opportunities for future military applications of advanced speech technology, and (3) identification of problem areas where research in speech processing is needed to meet application requirements, and of current research thrusts which appear promising. Opportunities for advanced applications are identified by means of descriptions of several generic systems which would be possible with advances in speech technology and in system integration. These generic systems include an integrated multirate voice/data communications terminal, an interactive speech enhancement system, a voice-controlled pilot's associate system, advanced air traffic control training systems, a battle management command and control support system with spoken natural language interface, and a spoken language translation system. In the applications discussed, various spectral analysis and digitization techniques are teamed up with artificial intelligence schemes to allow a machine to recognize continuous-speech sentences with vocabularies running to thousands of words. >