Improving automatic speaker verification using SVM techniques

Automatic recognition of people is an active field of research with important forensic and security applications. In these applications, it is not always possible for the subject to be in close proximity to the system. Voice represents a human behavioural trait which can be used to recognise people in such situations. Automatic Speaker Verification (ASV) is the process of verifying a persons identity through the analysis of their speech and enables recognition of a subject at a distance over a telephone channel { wired or wireless. A significant amount of research has focussed on the application of Gaussian mixture model (GMM) techniques to speaker verification systems providing state-of-the-art performance. GMM's are a type of generative classifier trained to model the probability distribution of the features used to represent a speaker. Recently introduced to the field of ASV research is the support vector machine (SVM). An SVM is a discriminative classifier requiring examples from both positive and negative classes to train a speaker model. The SVM is based on margin maximisation whereby a hyperplane attempts to separate classes in a high dimensional space. SVMs applied to the task of speaker verification have shown high potential, particularly when used to complement current GMM-based techniques in hybrid systems. This work aims to improve the performance of ASV systems using novel and innovative SVM-based techniques. Research was divided into three main themes: session variability compensation for SVMs; unsupervised model adaptation; and impostor dataset selection. The first theme investigated the differences between the GMM and SVM domains for the modelling of session variability | an aspect crucial for robust speaker verification. Techniques developed to improve the robustness of GMMbased classification were shown to bring about similar benefits to discriminative SVM classification through their integration in the hybrid GMM mean supervector SVM classifier. Further, the domains for the modelling of session variation were contrasted to find a number of common factors, however, the SVM-domain consistently provided marginally better session variation compensation. Minimal complementary information was found between the techniques due to the similarities in how they achieved their objectives. The second theme saw the proposal of a novel model for the purpose of session variation compensation in ASV systems. Continuous progressive model adaptation attempts to improve speaker models by retraining them after exploiting all encountered test utterances during normal use of the system. The introduction of the weight-based factor analysis model provided significant performance improvements of over 60% in an unsupervised scenario. SVM-based classification was then integrated into the progressive system providing further benefits in performance over the GMM counterpart. Analysis demonstrated that SVMs also hold several beneficial characteristics to the task of unsupervised model adaptation prompting further research in the area. In pursuing the final theme, an innovative background dataset selection technique was developed. This technique selects the most appropriate subset of examples from a large and diverse set of candidate impostor observations for use as the SVM background by exploiting the SVM training process. This selection was performed on a per-observation basis so as to overcome the shortcoming of the traditional heuristic-based approach to dataset selection. Results demonstrate the approach to provide performance improvements over both the use of the complete candidate dataset and the best heuristically-selected dataset whilst being only a fraction of the size. The refined dataset was also shown to generalise well to unseen corpora and be highly applicable to the selection of impostor cohorts required in alternate techniques for speaker verification.

Performing composition : developing a computer assisted composition system through live coding performance

This thesis maps the author's journey from a music composition practice to a composition and performance practice. The work involves the development of a software library for the purpose of encapsulating compositional ideas in software, and realising these ideas in performance through a live coding computer music practice. The thesis examines what artistic practice emerges through live coding and software development, and does this permit a blurring between the activities of music composition and performance. The role that software design plays in affecting musical outcomes is considered to gain an insight into how software development contributes to artistic development. The relationship between music composition and performance is also examined to identify the means by which engaging in live coding and software development can bring these activities together. The thesis, situated within the discourse of practice led research, documents a journey which uses the experience of software development and performance as a means to guide the direction of the research. The journey serves as an experiment for the author in engaging an hitherto unfamiliar musical practice, and as a roadmap for others seeking to modify or broaden their artistic practice.

Evolutionary algorithms for sustainable building design

This approach to sustainable design explores the possibility of creating an architectural design process which can iteratively produce optimised and sustainable design solutions. Driven by an evolution process based on genetic algorithms, the system allows the designer to “design the building design generator” rather than to “designs the building”. The design concept is abstracted into a digital design schema, which allows transfer of the human creative vision into the rational language of a computer. The schema is then elaborated into the use of genetic algorithms to evolve innovative, performative and sustainable design solutions. The prioritisation of the project’s constraints and the subsequent design solutions synthesised during design generation are expected to resolve most of the major conflicts in the evaluation and optimisation phases. Mosques are used as the example building typology to ground the research activity. The spatial organisations of various mosque typologies are graphically represented by adjacency constraints between spaces. Each configuration is represented by a planar graph which is then translated into a non-orthogonal dual graph and fed into the genetic algorithm system with fixed constraints and expected performance criteria set to govern evolution. The resultant Hierarchical Evolutionary Algorithmic Design System is developed by linking the evaluation process with environmental assessment tools to rank the candidate designs. The proposed system generates the concept, the seed, and the schema, and has environmental performance as one of the main criteria in driving optimisation.

Anticipatory timing in algorithmic rhythm generation

Generative music algorithms frequently operate by making musical decisions in a sequence, with each step of the sequence incorporating the local musical context in the decision process. The context is generally a short window of past musical actions. What is not generally included in the context is future actions. For real-time systems this is because the future is unknown. Offline systems also frequently utilise causal algorithms either for reasons of efficiency [1] or to simulate perceptual constraints [2]. However, even real-time agents can incorporate knowledge of their own future actions by utilising some form of planning. We argue that for rhythmic generation the incorporation of a limited form of planning - anticipatory timing - offers a worthwhile trade-off between musical salience and efficiency. We give an example of a real-time generative agent - the Jambot - that utilises anticipatory timing for rhythmic generation. We describe its operation, and compare its output with and without anticipatory timing.

Interrogating statistical models of music perception

We have developed a new experimental method for interrogating statistical theories of music perception by implementing these theories as generative music algorithms. We call this method Generation in Context. This method differs from most experimental techniques in music perception in that it incorporates aesthetic judgments. Generation In Context is designed to measure percepts for which the musical context is suspected to play an important role. In particular the method is suitable for the study of perceptual parameters which are temporally dynamic. We outline a use of this approach to investigate David Temperley’s (2007) probabilistic melody model, and provide some provisional insights as to what is revealed about the model. We suggest that Temperley’s model could be improved by dynamically modulating the probability distributions according to the changing musical context.

Navigating technological contexts and experience design in music education

For some time we have jokingly referred to our network jamming research with jam2jam as ‘Switched on Orff’ (Brown, Sorensen and Dillon 2002; Dillon 2003; Dillon 2006; Dillon 2006; Brown and Dillon 2007). The connection with electronic music and Wendy Carlos’ classic work ‘Switched on Bach’ was obvious; we were using electronic music in schools and with children. The deeper connection with Orff however was about recognising that electronic music and instruments could have cultural values and knowledge embedded in their design and practice in same way as what has come to be known as the Orff method (Orff and Keetman 1958-66). However whilst the Orff method focuses upon Western art music perceptual framework electronic instruments have the potential to have more fluid musical environments and even to move to interdisciplinary study by including visual media. Whilst the Orff method focused on making sense of Western art music through experience electronic environments potentially can make sense of the world of multi media that pervades our lives.

Access to meaningful relationships through virtual instruments and ensembles

This workshop focuses upon research about the qualities of community in music and of music in community facilitated by technologically supported relationships. Generative media systems present an opportunity for users to leverage computational systems to form new relationships through interactive and collaborative experiences. Generative music and art are a relatively new phenomenon that use procedural invention as a creative technique to produce music and visual media. Early systems have demonstrated the potential to provide access to collaborative ensemble experiences for users with little formal musical or artistic expertise. This workshop examines the relational affordances of these systems evidenced by selected field data drawn from the Network Jamming Project. These generative performance systems enable access to unique ensembles with very little musical knowledge or skill and offer the possibility of interactive relationships with artists and musical knowledge through collaborative performance. In this workshop we will focus on data that highlights how these simulated experiences might lead to understandings that may be of social benefit. Conference participants will be invited to jam in real time using virtual interfaces and to evaluate purposively selected video artifacts that demonstrate different kinds of interactive relationship with artists, peers, and community and that enrich the sense of expressive self. Theoretical insights about meaningful engagement drawn from the longitudinal and cross cultural experiences will underpin the discussion and practical presentation.

Patterns, heuristics for architectural design support: making use of evolutionary modelling in design

Software used by architectural and industrial designers – has moved from becoming a tool for drafting, towards use in verification, simulation, project management and project sharing remotely. In more advanced models, parameters for the designed object can be adjusted so a family of variations can be produced rapidly. With advances in computer aided design technology, numerous design options can now be generated and analyzed in real time. However the use of digital tools to support design as an activity is still at an early stage and has largely been limited in functionality with regard to the design process. To date, major CAD vendors have not developed an integrated tool that is able to both leverage specialized design knowledge from various discipline domains (known as expert knowledge systems) and support the creation of design alternatives that satisfy different forms of constraints. We propose that evolutionary computing and machine learning be linked with parametric design techniques to record and respond to a designer’s own way of working and design history. It is expected that this will lead to results that impact on future work on design support systems-(ergonomics and interface) as well as implicit constraint and problem definition for problems that are difficult to quantify.

A statistical framework for natural feature representation

This paper presents a robust stochastic framework for the incorporation of visual observations into conventional estimation, data fusion, navigation and control algorithms. The representation combines Isomap, a non-linear dimensionality reduction algorithm, with expectation maximization, a statistical learning scheme. The joint probability distribution of this representation is computed offline based on existing training data. The training phase of the algorithm results in a nonlinear and non-Gaussian likelihood model of natural features conditioned on the underlying visual states. This generative model can be used online to instantiate likelihoods corresponding to observed visual features in real-time. The instantiated likelihoods are expressed as a Gaussian mixture model and are conveniently integrated within existing non-linear filtering algorithms. Example applications based on real visual data from heterogenous, unstructured environments demonstrate the versatility of the generative models.

A Bayesian approach for place recognition

This paper presents a robust place recognition algorithm for mobile robots. The framework proposed combines nonlinear dimensionality reduction, nonlinear regression under noise, and variational Bayesian learning to create consistent probabilistic representations of places from images. These generative models are learnt from a few images and used for multi-class place recognition where classification is computed from a set of feature-vectors. Recognition can be performed in near real-time and accounts for complexity such as changes in illumination, occlusions and blurring. The algorithm was tested with a mobile robot in indoor and outdoor environments with sequences of 1579 and 3820 images respectively. This framework has several potential applications such as map building, autonomous navigation, search-rescue tasks and context recognition.

Shifting the technology context : career-change entrants’ transition into teaching

Becoming a teacher in technology-rich classrooms is a complex and challenging transition for career-change entrants. Those with generic or specialist Information and Communication Technology (ICT) expertise bring a mindset about purposeful uses of ICT that enrich student learning and school communities. The transition process from a non-education environment is both enhanced and constrained by shifting the technology context of generic or specialist ICT expertise, developed through a former career as well as general life experience. In developing an understanding of the complexity of classrooms and creating a learner centred way of working, perceptions about learners and learning evolve and shift. Shifts in thinking about how ICT expertise supports learners and enhances learning preceded shifts in perceptions about being a teacher, working with colleagues, and functioning in schools that have varying degrees of intensity and impact on evolving professional identities. Current teacher education and school induction programs are seen to be falling short of meeting the needs of career-change entrants and, as a flow on, the students they nurture. Research (see, for example, Tigchelaar, Brouwer, & Korthagen, 2008; Williams & Forgasz, 2009) highlights the value of generic and specialist expertise career-change teachers bring to the profession and draws attention to the challenges such expertise begets (Anthony & Ord, 2008; Priyadharshini & Robinson-Pant, 2003). As such, the study described in this thesis investigated perceptions of career-change entrants, who have generic (Mishra & Koehler, 2006) or specialist expertise, that is, ICT qualifications and work experience in the use of ICT. The career-change entrants‘ perceptions were sought as they shifted the technology context and transitioned into teaching in technology-rich classrooms. The research involved an interpretive analysis of qualitative data and quantitative data. The study used the explanatory case study (Yin, 1994) methodology enriched through grounded theory processes (Strauss & Corbin, 1998), to develop a theory about professional identity transition from the perceptions of the participants in the study. The study provided insights into the expertise and experiences of career change entrants, particularly in relation to how professional identities that include generic and specialist ICT knowledge and expertise were reconfigured while transitioning into the teaching profession. This thesis presents the Professional Identity Transition Theory that encapsulates perceptions about teaching in technology-rich classrooms amongst a selection of the increasing number of career-change entrants. The theory, grounded in the data, (Strauss & Corbin, 1998) proposes that career-change entrants experience transition phases of varying intensity that impact on professional identity, retention and development as a teacher. These phases are linked to a shift in perceptions rather than time as a teacher. Generic and specialist expertise in the use of ICT is a weight of the past and an asset that makes the transition process more challenging for career-change entrants. The study showed that career-change entrants used their experiences and perceptions to develop a way of working in a school community. Their way of working initially had an adaptive orientation focussed on immediate needs as their teaching practice developed. Following a shift of thinking, more generative ways of working focussed on the future emerged to enable continual enhancement and development of practice. Sustaining such learning is a personal, school and systemic challenge for the teaching profession.

Note sequence morphing algorithms for performance of electronic dance music

This paper describes algorithms that can musically augment the realtime performance of electronic dance music by generating new musical material by morphing. Note sequence morphing involves the algorithmic generation of music that smoothly transitions between two existing musical segments. The potential of musical morphing in electronic dance music is outlined and previous research is summarised; including discussions of relevant music theoretic and algorithmic concepts. An outline and explanation is provided of a novel Markov morphing process that uses similarity measures to construct transition matrices. The paper reports on a ‘focus-concert’ study used to evaluate this morphing algorithm and to compare its output with performances from a professional DJ. Discussions of this trial include reflections on some of the aesthetic characteristics of note sequence morphing. The research suggests that the proposed morphing technique could be effectively used in some electronic dance music contexts.

Using text-spotting to query the world

The world we live in is well labeled for the benefit of humans but to date robots have made little use of this resource. In this paper we describe a system that allows robots to read and interpret visible text and use it to understand the content of the scene. We use a generative probabilistic model that explains spotted text in terms of arbitrary search terms. This allows the robot to understand the underlying function of the scene it is looking at, such as whether it is a bank or a restaurant. We describe the text spotting engine at the heart of our system that is able to detect and parse wild text in images, and the generative model, and present results from images obtained with a robot in a busy city setting.

Introducing action research as a vehicle to bring about change to a psychiatric nursing curriculum in Taiwan

The focus of this study is on curriculum change within a School of Nursing in Taiwan where there is a growing demand for educational reform in order to meet the new accreditation standards and demands of the Taiwan Nursing Accreditation Council (TNAC). The aim of this study was to transform the Psychiatric Nursing curriculum in ways that are empowering, generative and sustainable. This study introduced Action Research as a vehicle to bring about curriculum transformation. I conceptualised a framework to guide the transformation process based on the notions of learner-centredness, conceptual change, pedagogical knowledge, reflection, collaboration, reculturing and empowerment. The Action Plan was developed in accordance with the conceptual framework, and was developed in five steps through which team members explored and became aware of our conceptions of teaching and learning, and then planned and implemented actions to change our curriculum, and examined and reflected on the curriculum transformation. The study demonstrated the value of working collaboratively to solve educational problems. This study also suggested that experiential knowledge, when shared and integrated with theoretical knowledge, can constructively contribute to all aspects of curriculum transformation. This study further supported the value of including clinical facilitators in the development and transformation of curricula. It confirmed that academics and clinical facilitators can work together to create new learning for students. This study is significant for both practical and political reasons. Its practical significance lies in its direct utility to the learners and teachers who were involved in the study. The political significance lies in the potential of the study to lead to further changes or improvements in other, similar contexts. The study is limited in that any interpretations cannot be generalised to other contexts. However, what emerged adds to the body of knowledge in such a way that it would constitute the basis for better informed educational practice.

Lumia : Art | light | motion. The Art of Kuuki.

The State Library of Queensland is delighted to present Lumia: art/light/motion, a culmination of many years of collaboration by the Kuuki collective led by Priscilla Bracks and Gavin Sade. This extraordinary exhibition not only showcases the unique talent of these Queenslanders, it also opens up a world of future possibilities while re-presenting the past and present. These contemporary new media installations sit comfortably within the walls of the library as they are the distinctive products of inquisitive and philosophical minds. In a sense the exhibition highlights the longevity and purposefulness of a cultural learning institution, through the non-traditional use of data, information, research and collection interpretation. The exhibition simultaneously articulates one of our key objectives – to progress the state’s digital agenda. Two academic essays have been commissioned for this joint Kuuki and State Library of Queensland publication. The first is by artist and writer Paul Brown, who has specialised in art, science and technology since the late 1960s and in computational and generative art since the mid 1970s. Brown investigates the history of new media, which is celebrating its 60th anniversary, and clearly places Sade and Bracks at the forefront of this genre nationally. The second essay is by arts writer Linda Carroli, who has delved deeply into the thoughts and processes of the artists to bring to light the complex workings of the artists’ minds. The publication also features an interview Carroli conducted with the artists. This exhibition is playful, informative and contemplative. The audience is invited to play, and consequently to ponder the way we live and the environmental and social implications of our choices. The exhibition tempts us to travel deep into the Antarctic, plunge into the Great Barrier Reef, be swamped by an orchestra of crickets, enter the Charmed world and travel back in time to a Victorian parlour where you can interact with a ‘new-world’ lyrebird and consider a brave new world where our only link to the animal world is with robotic representations. In essence this exhibition is about ideas and knowledge and what better institution than the State Library of Queensland to partner such a project?. State Library is committed to preserving culture, exploring new media and creating new content as a lasting legacy of Queensland for all Queenslanders.