Multi Pattern Dynamic Time Warping For Automatic Speech Recognition

We are addressing the problem of jointly using multiple noisy speech patterns for automatic speech recognition (ASR), given that they come from the same class. If the user utters a word K times, the ASR system should try to use the information content in all the K patterns of the word simultaneously and improve its speech recognition accuracy compared to that of the single pattern based speech recognition. T address this problem, recently we proposed a Multi Pattern Dynamic Time Warping (MPDTW) algorithm to align the K patterns by finding the least distortion path between them. A Constrained Multi Pattern Viterbi algorithm was used on this aligned path for isolated word recognition (IWR). In this paper, we explore the possibility of using only the MPDTW algorithm for IWR. We also study the properties of the MPDTW algorithm. We show that using only 2 noisy test patterns (10 percent burst noise at -5 dB SNR) reduces the noisy speech recognition error rate by 37.66 percent when compared to the single pattern recognition using the Dynamic Time Warping algorithm.

Cortical processing of speech and non-speech sounds in adults and newborns

Comprehension of a complex acoustic signal - speech - is vital for human communication, with numerous brain processes required to convert the acoustics into an intelligible message. In four studies in the present thesis, cortical correlates for different stages of speech processing in a mature linguistic system of adults were investigated. In two further studies, developmental aspects of cortical specialisation and its plasticity in adults were examined. In the present studies, electroencephalographic (EEG) and magnetoencephalographic (MEG) recordings of the mismatch negativity (MMN) response elicited by changes in repetitive unattended auditory events and the phonological mismatch negativity (PMN) response elicited by unexpected speech sounds in attended speech inputs served as the main indicators of cortical processes. Changes in speech sounds elicited the MMNm, the magnetic equivalent of the electric MMN, that differed in generator loci and strength from those elicited by comparable changes in non-speech sounds, suggesting intra- and interhemispheric specialisation in the processing of speech and non-speech sounds at an early automatic processing level. This neuronal specialisation for the mother tongue was also reflected in the more efficient formation of stimulus representations in auditory sensory memory for typical native-language speech sounds compared with those formed for unfamiliar, non-prototype speech sounds and simple tones. Further, adding a speech or non-speech sound context to syllable changes was found to modulate the MMNm strength differently in the left and right hemispheres. Following the acoustic-phonetic processing of speech input, phonological effort related to the selection of possible lexical (word) candidates was linked with distinct left-hemisphere neuronal populations. In summary, the results suggest functional specialisation in the neuronal substrates underlying different levels of speech processing. Subsequently, plasticity of the brain's mature linguistic system was investigated in adults, in whom representations for an aurally-mediated communication system, Morse code, were found to develop within the same hemisphere where representations for the native-language speech sounds were already located. Finally, recording and localization of the MMNm response to changes in speech sounds was successfully accomplished in newborn infants, encouraging future MEG investigations on, for example, the state of neuronal specialisation at birth.

Cortical processing of speech and non-speech sounds in autism and Asperger syndrome

Autism and Asperger syndrome (AS) are neurodevelopmental disorders characterised by deficient social and communication skills, as well as restricted, repetitive patterns of behaviour. The language development in individuals with autism is significantly delayed and deficient, whereas in individuals with AS, the structural aspects of language develop quite normally. Both groups, however, have semantic-pragmatic language deficits. The present thesis investigated auditory processing in individuals with autism and AS. In particular, the discrimination of and orienting to speech and non-speech sounds was studied, as well as the abstraction of invariant sound features from speech-sound input. Altogether five studies were conducted with auditory event-related brain potentials (ERP); two studies also included a behavioural sound-identification task. In three studies, the subjects were children with autism, in one study children with AS, and in one study adults with AS. In children with autism, even the early stages of sound encoding were deficient. In addition, these children had altered sound-discrimination processes characterised by enhanced spectral but deficient temporal discrimination. The enhanced pitch discrimination may partly explain the auditory hypersensitivity common in autism, and it may compromise the filtering of relevant auditory information from irrelevant information. Indeed, it was found that when sound discrimination required abstracting invariant features from varying input, children with autism maintained their superiority in pitch processing, but lost it in vowel processing. Finally, involuntary orienting to sound changes was deficient in children with autism in particular with respect to speech sounds. This finding is in agreement with previous studies on autism suggesting deficits in orienting to socially relevant stimuli. In contrast to children with autism, the early stages of sound encoding were fairly unimpaired in children with AS. However, sound discrimination and orienting were rather similarly altered in these children as in those with autism, suggesting correspondences in the auditory phenotype in these two disorders which belong to the same continuum. Unlike children with AS, adults with AS showed enhanced processing of duration changes, suggesting developmental changes in auditory processing in this disorder.

Porn, censorship, classification and free speech: Global paradoxes in the governance of media content

In this presentation, I reflect upon the global landscape surrounding the governance and classification of media content, at a time of rapid change in media platforms and services for content production and distribution, and contested cultural and social norms. I discuss the tensions and contradictions arising in the relationship between national, regional and global dimensions of media content distribution, as well as the changing relationships between state and non-state actors. These issues will be explored through consideration of issues such as: recent debates over film censorship; the review of the National Classification Scheme conducted by the Australian Law Reform Commission; online controversies such as the future of the Reddit social media site; and videos posted online by the militant group ISIS.

Audience members watching the opening ceremonies of the Heidelberg festival, with opening speech given by Thomas Mann.

Typed caption: Festakt in der Stadthalle zur Eroeffnung der Heidelberger Festpiele 1929. Thomas Mann hielt die Festrede. Unsere Aufnahme zeigt einen Ausschnitt aus der Festversammlung, an der die geistige Prominenz Heidelbergs und zahlreiche bedeutende Persoenlichkeiten des deutschen Kulturlebends teilnahmen. Von links nach rechts in der 1. Reihe: Rene Schickele, Rudolf Rittner, Gustav Hargung, Prof. Dr. Martin Dibelius, Gerhart Hauptmann, Dr. Rudolf H. Goldschmidt, Oberbuergermeister Dr. Neinhaus, Thomas Mann und Frau, Kultusminister Leers, Rudolf G. Binding; In der 3. Reihe: Oberbaurat L. Schmieder, Nobelpreistraeger Dr. F. Bergius, Bankdirektor Fremerey, Prof. Hellpach, Prof. Radbruch

Corporate Social Responsibility (CSR) in junior and mid-tier Australian resources companies operating in developing nations - beyond the public relations offensive

NIt is now widely accepted that corporations have a responsibility to benefit society, as well as generate profit. This study used institutional theory to explore how the complex and contested notion of corporate social responsibility is understood and practiced by junior and mid-tier Australian resources companies operating in the world's most impoverished countries. The study found that CSR meaning and practice in this large but little researched group of companies was shaped by complex pressures at the global, industry, organisational and individual levels. Importantly, the study also revealed striking contradictions and ambiguities between participants' CSR aspirations and their actions and accountability.

Compressive Sensing For Sparsely Excited Speech Signals

Compressive sensing (CS) has been proposed for signals with sparsity in a linear transform domain. We explore a signal dependent unknown linear transform, namely the impulse response matrix operating on a sparse excitation, as in the linear model of speech production, for recovering compressive sensed speech. Since the linear transform is signal dependent and unknown, unlike the standard CS formulation, a codebook of transfer functions is proposed in a matching pursuit (MP) framework for CS recovery. It is found that MP is efficient and effective to recover CS encoded speech as well as jointly estimate the linear model. Moderate number of CS measurements and low order sparsity estimate will result in MP converge to the same linear transform as direct VQ of the LP vector derived from the original signal. There is also high positive correlation between signal domain approximation and CS measurement domain approximation for a large variety of speech spectra.

Robust and High-resolution Voiced/Unvoiced Classification in Noisy Speech Using A Signal Smoothness Criterion

We propose a novel technique for robust voiced/unvoiced segment detection in noisy speech, based on local polynomial regression. The local polynomial model is well-suited for voiced segments in speech. The unvoiced segments are noise-like and do not exhibit any smooth structure. This property of smoothness is used for devising a new metric called the variance ratio metric, which, after thresholding, indicates the voiced/unvoiced boundaries with 75% accuracy for 0dB global signal-to-noise ratio (SNR). A novelty of our algorithm is that it processes the signal continuously, sample-by-sample rather than frame-by-frame. Simulation results on TIMIT speech database (downsampled to 8kHz) for various SNRs are presented to illustrate the performance of the new algorithm. Results indicate that the algorithm is robust even in high noise levels.

Two Stage Transform Vector Quantization of LSFs for Wideband Speech Coding

We investigate the use of a two stage transform vector quantizer (TSTVQ) for coding of line spectral frequency (LSF) parameters in wideband speech coding. The first stage quantizer of TSTVQ, provides better matching of source distribution and the second stage quantizer provides additional coding gain through using an individual cluster specific decorrelating transform and variance normalization. Further coding gain is shown to be achieved by exploiting the slow time-varying nature of speech spectra and thus using inter-frame cluster continuity (ICC) property in the first stage of TSTVQ method. The proposed method saves 3-4 bits and reduces the computational complexity by 58-66%, compared to the traditional split vector quantizer (SVQ), but at the expense of 1.5-2.5 times of memory.

Joint evaluation of multiple speech patterns for speech recognition and training

We are addressing the novel problem of jointly evaluating multiple speech patterns for automatic speech recognition and training. We propose solutions based on both the non-parametric dynamic time warping (DTW) algorithm, and the parametric hidden Markov model (HMM). We show that a hybrid approach is quite effective for the application of noisy speech recognition. We extend the concept to HMM training wherein some patterns may be noisy or distorted. Utilizing the concept of ``virtual pattern'' developed for joint evaluation, we propose selective iterative training of HMMs. Evaluating these algorithms for burst/transient noisy speech and isolated word recognition, significant improvement in recognition accuracy is obtained using the new algorithms over those which do not utilize the joint evaluation strategy.

Seeing and hearing speech, sounds, and signs: functional magnetic resonance studies on fluent and dyslexic readers

Speech has both auditory and visual components (heard speech sounds and seen articulatory gestures). During all perception, selective attention facilitates efficient information processing and enables concentration on high-priority stimuli. Auditory and visual sensory systems interact at multiple processing levels during speech perception and, further, the classical motor speech regions seem also to participate in speech perception. Auditory, visual, and motor-articulatory processes may thus work in parallel during speech perception, their use possibly depending on the information available and the individual characteristics of the observer. Because of their subtle speech perception difficulties possibly stemming from disturbances at elemental levels of sensory processing, dyslexic readers may rely more on motor-articulatory speech perception strategies than do fluent readers. This thesis aimed to investigate the neural mechanisms of speech perception and selective attention in fluent and dyslexic readers. We conducted four functional magnetic resonance imaging experiments, during which subjects perceived articulatory gestures, speech sounds, and other auditory and visual stimuli. Gradient echo-planar images depicting blood oxygenation level-dependent contrast were acquired during stimulus presentation to indirectly measure brain hemodynamic activation. Lip-reading activated the primary auditory cortex, and selective attention to visual speech gestures enhanced activity within the left secondary auditory cortex. Attention to non-speech sounds enhanced auditory cortex activity bilaterally; this effect showed modulation by sound presentation rate. A comparison between fluent and dyslexic readers' brain hemodynamic activity during audiovisual speech perception revealed stronger activation of predominantly motor speech areas in dyslexic readers during a contrast test that allowed exploration of the processing of phonetic features extracted from auditory and visual speech. The results show that visual speech perception modulates hemodynamic activity within auditory cortex areas once considered unimodal, and suggest that the left secondary auditory cortex specifically participates in extracting the linguistic content of seen articulatory gestures. They are strong evidence for the importance of attention as a modulator of auditory cortex function during both sound processing and visual speech perception, and point out the nature of attention as an interactive process (influenced by stimulus-driven effects). Further, they suggest heightened reliance on motor-articulatory and visual speech perception strategies among dyslexic readers, possibly compensating for their auditory speech perception difficulties.

Joint decoding of multiple speech patterns for robust speech recognition

We are addressing a new problem of improving automatic speech recognition performance, given multiple utterances of patterns from the same class. We have formulated the problem of jointly decoding K multiple patterns given a single Hidden Markov Model. It is shown that such a solution is possible by aligning the K patterns using the proposed Multi Pattern Dynamic Time Warping algorithm followed by the Constrained Multi Pattern Viterbi Algorithm The new formulation is tested in the context of speaker independent isolated word recognition for both clean and noisy patterns. When 10 percent of speech is affected by a burst noise at -5 dB Signal to Noise Ratio (local), it is shown that joint decoding using only two noisy patterns reduces the noisy speech recognition error rate to about 51 percent, when compared to the single pattern decoding using the Viterbi Algorithm. In contrast a simple maximization of individual pattern likelihoods, provides only about 7 percent reduction in error rate.

GMM based Bayesian approach to speech enhancement in signal transform domain

Considering a general linear model of signal degradation, by modeling the probability density function (PDF) of the clean signal using a Gaussian mixture model (GMM) and additive noise by a Gaussian PDF, we derive the minimum mean square error (MMSE) estimator. The derived MMSE estimator is non-linear and the linear MMSE estimator is shown to be a special case. For speech signal corrupted by independent additive noise, by modeling the joint PDF of time-domain speech samples of a speech frame using a GMM, we propose a speech enhancement method based on the derived MMSE estimator. We also show that the same estimator can be used for transform-domain speech enhancement.