FPGA implementation of dual-microphone delay-and-sum beamforming for in-car speech enhancement and recognition

In an automotive environment, the performance of a speech recognition system is affected by environmental noise if the speech signal is acquired directly from a microphone. Speech enhancement techniques are therefore necessary to improve the speech recognition performance. In this paper, a field-programmable gate array (FPGA) implementation of dual-microphone delay-and-sum beamforming (DASB) for speech enhancement is presented. As the first step towards a cost-effective solution, the implementation described in this paper uses a relatively high-end FPGA device to facilitate the verification of various design strategies and parameters. Experimental results show that the proposed design can produce output waveforms close to those generated by a theoretical (floating-point) model with modest usage of FPGA resources. Speech recognition experiments are also conducted on enhanced in-car speech waveforms produced by the FPGA in order to compare recognition performance with the floating-point representation running on a PC.

Exploring visual features through Gabor representations for facial expression detection

Gabor representations have been widely used in facial analysis (face recognition, face detection and facial expression detection) due to their biological relevance and computational properties. Two popular Gabor representations used in literature are: 1) Log-Gabor and 2) Gabor energy filters. Even though these representations are somewhat similar, they also have distinct differences as the Log-Gabor filters mimic the simple cells in the visual cortex while the Gabor energy filters emulate the complex cells, which causes subtle differences in the responses. In this paper, we analyze the difference between these two Gabor representations and quantify these differences on the task of facial action unit (AU) detection. In our experiments conducted on the Cohn-Kanade dataset, we report an average area underneath the ROC curve (A`) of 92.60% across 17 AUs for the Gabor energy filters, while the Log-Gabor representation achieved an average A` of 96.11%. This result suggests that small spatial differences that the Log-Gabor filters pick up on are more useful for AU detection than the differences in contours and edges that the Gabor energy filters extract.

Acoustic keyword spotting in speech with applications to data mining

Keyword Spotting is the task of detecting keywords of interest within continu- ous speech. The applications of this technology range from call centre dialogue systems to covert speech surveillance devices. Keyword spotting is particularly well suited to data mining tasks such as real-time keyword monitoring and unre- stricted vocabulary audio document indexing. However, to date, many keyword spotting approaches have su®ered from poor detection rates, high false alarm rates, or slow execution times, thus reducing their commercial viability. This work investigates the application of keyword spotting to data mining tasks. The thesis makes a number of major contributions to the ¯eld of keyword spotting. The ¯rst major contribution is the development of a novel keyword veri¯cation method named Cohort Word Veri¯cation. This method combines high level lin- guistic information with cohort-based veri¯cation techniques to obtain dramatic improvements in veri¯cation performance, in particular for the problematic short duration target word class. The second major contribution is the development of a novel audio document indexing technique named Dynamic Match Lattice Spotting. This technique aug- ments lattice-based audio indexing principles with dynamic sequence matching techniques to provide robustness to erroneous lattice realisations. The resulting algorithm obtains signi¯cant improvement in detection rate over lattice-based audio document indexing while still maintaining extremely fast search speeds. The third major contribution is the study of multiple veri¯er fusion for the task of keyword veri¯cation. The reported experiments demonstrate that substantial improvements in veri¯cation performance can be obtained through the fusion of multiple keyword veri¯ers. The research focuses on combinations of speech background model based veri¯ers and cohort word veri¯ers. The ¯nal major contribution is a comprehensive study of the e®ects of limited training data for keyword spotting. This study is performed with consideration as to how these e®ects impact the immediate development and deployment of speech technologies for non-English languages.

Automatic spoken language identification utilizing acoustic and phonetic speech information

Automatic spoken Language Identi¯cation (LID) is the process of identifying the language spoken within an utterance. The challenge that this task presents is that no prior information is available indicating the content of the utterance or the identity of the speaker. The trend of globalization and the pervasive popularity of the Internet will amplify the need for the capabilities spoken language identi¯ca- tion systems provide. A prominent application arises in call centers dealing with speakers speaking di®erent languages. Another important application is to index or search huge speech data archives and corpora that contain multiple languages. The aim of this research is to develop techniques targeted at producing a fast and more accurate automatic spoken LID system compared to the previous National Institute of Standards and Technology (NIST) Language Recognition Evaluation. Acoustic and phonetic speech information are targeted as the most suitable fea- tures for representing the characteristics of a language. To model the acoustic speech features a Gaussian Mixture Model based approach is employed. Pho- netic speech information is extracted using existing speech recognition technol- ogy. Various techniques to improve LID accuracy are also studied. One approach examined is the employment of Vocal Tract Length Normalization to reduce the speech variation caused by di®erent speakers. A linear data fusion technique is adopted to combine the various aspects of information extracted from speech. As a result of this research, a LID system was implemented and presented for evaluation in the 2003 Language Recognition Evaluation conducted by the NIST.

Semi-supervised intelligent surveillance system for secure environments

This paper proposes a semi-supervised intelligent visual surveillance system to exploit the information from multi-camera networks for the monitoring of people and vehicles. Modules are proposed to perform critical surveillance tasks including: the management and calibration of cameras within a multi-camera network; tracking of objects across multiple views; recognition of people utilising biometrics and in particular soft-biometrics; the monitoring of crowds; and activity recognition. Recent advances in these computer vision modules and capability gaps in surveillance technology are also highlighted.

The use of temporal speech and lip information for multi-modal speaker identification via multi-stream HMMs

Investigates the use of temporal lip information, in conjunction with speech information, for robust, text-dependent speaker identification. We propose that significant speaker-dependent information can be obtained from moving lips, enabling speaker recognition systems to be highly robust in the presence of noise. The fusion structure for the audio and visual information is based around the use of multi-stream hidden Markov models (MSHMM), with audio and visual features forming two independent data streams. Recent work with multi-modal MSHMMs has been performed successfully for the task of speech recognition. The use of temporal lip information for speaker identification has been performed previously (T.J. Wark et al., 1998), however this has been restricted to output fusion via single-stream HMMs. We present an extension to this previous work, and show that a MSHMM is a valid structure for multi-modal speaker identification

Robust recognition of human behaviour in challenging environments

Novel techniques have been developed for the automatic recognition of human behaviour in challenging environments using information from visual and infra-red camera feeds. The techniques have been applied to two interesting scenarios: Recognise drivers' speech using lip movements and recognising audience behaviour, while watching a movie, using facial features and body movements. Outcome of the research in these two areas will be useful in the improving the performance of voice recognition in automobiles for voice based control and for obtaining accurate movie interest ratings based on live audience response analysis.

Probabilistic orthographic cues to grammatical category in the brain

What helps us determine whether a word is a noun or a verb, without conscious awareness? We report on cues in the way individual English words are spelled, and, for the first time, identify their neural correlates via functional magnetic resonance imaging (fMRI). We used a lexical decision task with trisyllabic nouns and verbs containing orthographic cues that are either consistent or inconsistent with the spelling patterns of words from that grammatical category. Significant linear increases in response times and error rates were observed as orthography became less consistent, paralleled by significant linear decreases in blood oxygen level dependent (BOLD) signal in the left supramarginal gyrus of the left inferior parietal lobule, a brain region implicated in visual word recognition. A similar pattern was observed in the left superior parietal lobule. These findings align with an emergentist view of grammatical category processing which results from sensitivity to multiple probabilistic cues.

Acoustic adaptation in cross database audio visual SHMM training for phonetic spoken term detection

Visual information in the form of lip movements of the speaker has been shown to improve the performance of speech recognition and search applications. In our previous work, we proposed cross database training of synchronous hidden Markov models (SHMMs) to make use of external large and publicly available audio databases in addition to the relatively small given audio visual database. In this work, the cross database training approach is improved by performing an additional audio adaptation step, which enables audio visual SHMMs to benefit from audio observations of the external audio models before adding visual modality to them. The proposed approach outperforms the baseline cross database training approach in clean and noisy environments in terms of phone recognition accuracy as well as spoken term detection (STD) accuracy.

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.

Multiple Perspectives on Networks: Conceptual Development, Application and Integration in an Entrepreneurial Context

Multiple Perspectives on Networks: Conceptual Development, Application and Integration in an Entrepreneurial Context. The purpose of this thesis is to enhance cross-fertilization between three different approaches to network research. The business network approach may contribute in terms of how relationships are created, developed and how tie content changes within ties, not only between them. The social network approach adds to the discussion by offering concepts of structural change on a network level. The network approach in entrepreneurship contributes by emphasizing network content, governance and structure as a way of understanding and capturing networks. This is discussed in the conceptual articles, Articles 2 and 3. The ultimate purpose of this thesis is to develop a theoretical and empirical understanding of network development processes. This is fulfilled by presenting a theoretical framework, which offers multiple views on process as a developmental outcome. The framework implies that change ought to be captured both within and among relationships over time in the firm as well as in the network. Consequently, changes in structure and interaction taking place simultaneously need to be included when doing research on network development. The connection between micro and macro levels is also stressed. Therefore, the entrepreneur or firm level needs to be implemented together with the network level. The surrounding environment impacts firm and network development and vice versa and hence needs to be integrated. Further, it is necessary to view network development not only as a way forward but to include both progression and regression as inevitable parts of the process. Finally, both stability and change should be taken into account as part of network development. Empirical results in Article 1 show support for a positive impact of networks on SME internationalization. Article 4 compares networks of novice, serial and portfolio entrepreneurs but the empirical results show little support for differences in the networks by type of entrepreneur. The results demonstrate that network interaction and structure is not directly impacted by type of entrepreneur involved. It indicates instead that network structure and interaction is more impacted by the development phase of the firm. This in turn is in line with the theoretical implications, stating that the development of the network and the firm impacts each other, as they co-evolve.

Constructing a wire-frame from views on arbitrary view planes for objects with conic sections inclined to all view planes

A new and efficient approach to construct a 3D wire-frame of an object from its orthographic projections is described. The input projections can be two or more and can include regular and complete auxiliary views. Each view may contain linear, circular and other conic sections. The output is a 3D wire-frame that is consistent with the input views. The approach can handle auxiliary views containing curved edges. This generality derives from a new technique to construct 3D vertices from the input 2D vertices (as opposed to matching coordinates that is prevalent in current art). 3D vertices are constructed by projecting the 2D vertices in a pair of views on the common line of the two views. The construction of 3D edges also does not require the addition of silhouette and tangential vertices and subsequently splitting edges in the views. The concepts of complete edges and n-tuples are introduced to obviate this need. Entities corresponding to the 3D edge in each view are first identified and the 3D edges are then constructed from the information available with the matching 2D edges. This allows the algorithm to handle conic sections that are not parallel to any of the viewing directions. The localization of effort in constructing 3D edges is the source of efficiency of the construction algorithm as it does not process all potential 3D edges. Working of the algorithm on typical drawings is illustrated. (C) 2011 Elsevier Ltd. All rights reserved.

Tandem system adaptation using multiple linear feature transforms

Adaptation to speaker and environment changes is an essential part of current automatic speech recognition (ASR) systems. In recent years the use of multi-layer percpetrons (MLPs) has become increasingly common in ASR systems. A standard approach to handling speaker differences when using MLPs is to apply a global speaker-specific constrained MLLR (CMLLR) transform to the features prior to training or using the MLP. This paper considers the situation when there are both speaker and channel, communication link, differences in the data. A more powerful transform, front-end CMLLR (FE-CMLLR), is applied to the inputs to the MLP to represent the channel differences. Though global, these FE-CMLLR transforms vary from time-instance to time-instance. Experiments on a channel distorted dialect Arabic conversational speech recognition task indicates the usefulness of adapting MLP features using both CMLLR and FE-CMLLR transforms. © 2013 IEEE.

crossmodal error correction of continuous handwriting recognition by speech

In recognition-based user interface, users’ satisfaction is determined not only by recognition accuracy but also by effort to correct recognition errors. In this paper, we introduce a crossmodal error correction technique, which allows users to correct errors of Chinese handwriting recognition by speech. The focus of the paper is a multimodal fusion algorithm supporting the crossmodal error correction. By fusing handwriting and speech recognition, the algorithm can correct errors in both character extraction and recognition of handwriting. The experimental result indicates that the algorithm is effective and efficient. Moreover, the evaluation also shows the correction technique can help users to correct errors in handwriting recognition more efficiently than the other two error correction techniques.