Speaker attribution of multiple telephone conversations using a complete-linkage clustering approach

In this paper we propose and evaluate a speaker attribution system using a complete-linkage clustering method. Speaker attribution refers to the annotation of a collection of spoken audio based on speaker identities. This can be achieved using diarization and speaker linking. The main challenge associated with attribution is achieving computational efficiency when dealing with large audio archives. Traditional agglomerative clustering methods with model merging and retraining are not feasible for this purpose. This has motivated the use of linkage clustering methods without retraining. We first propose a diarization system using complete-linkage clustering and show that it outperforms traditional agglomerative and single-linkage clustering based diarization systems with a relative improvement of 40% and 68%, respectively. We then propose a complete-linkage speaker linking system to achieve attribution and demonstrate a 26% relative improvement in attribution error rate (AER) over the single-linkage speaker linking approach.

A review on environmental durability of CFRP strengthened system

Recently the use of the carbon fibre reinforced polymer(CFRP) composites appears to be an excellent solution for retrofitting and strengthening of concrete and steel structures because of its superior physical and mechanical properties through the integration of other materials. However, the overall functionality and durability under various environmental conditions of the system has not yet been well documented. This paper reviews the environmental durability of CFRP strengthened system that has received only small coverage in previous review articles. Future research topics have also been indentified, such as durability of steel circular hollow section under various environmental conditions subjected to bending. Environment of interests are moisture/solution, alkalinity, creep/relaxation, fatigue, fire, thermal effects (including freeze-thaw), and ultraviolet exposure.

Superconducting joints of Bi-2212/Ag tapes and laminates

Different types of HTS joints of Bi-2212/Ag tapes and laminates, which are fabricated by dip-coating and partial-melt processes, have been investigated. All joints are prepared using green single and laminated tapes and according to the scheme: coating-joining-processing. The heat treated tapes have critical current (Ic) between 7 and 27 A, depending on tape thickness and the number of Bi-2212 ceramic layers in laminated tapes. It is found that the current transport properties of joints depend on the type of laminate, joint configuration and joint treatment, Ic losses in joints of Bi-2212 tapes and laminates are attributed to defects in their structure, such as pores, secondary phases and misalignment of Bi-2212 grains near the Ag edges. By optimizing joint configuration, current transmission up to 100% is achieved for both single tapes and laminated tapes.

Abstraction as noise : enhancing the poetic

Abstraction in its resistance to evident meaning has the capacity to interrupt or at least provide tools with which to question an overly compliant reception of the information to which we are subject. It does so by highlighting a latency or potentiality inherent in materiality that points to the possibility of a critical resistance to this ceaseless flow of sound/image/data. This resistance has been remarked on in differing ways by a number of commentators such as Lyotard, in his exploration of the avant-garde and the sublime for example. This joint paper will initially map the collaborative project by Daniel Mafe and Andrew Brown, Affecting Interference which conjoins painting with digital sound and animations into a single, large scale, immersive exhibition/installation. The work acts as an interstitial point between contrasting approaches to abstraction: the visual and aural, the digital and analogue. The paper will then explore the ramifications of this through the examination of abstraction as ‘noise’, that is as that raw inassimilable materiality, within which lays the creative possibility to forge and embrace the as-yet-unthought and almost-forgotten. It does so by establishing a space for a more poetic and slower paced critical engagement for the viewing and receiving information or data. This slowing of perception through the suspension of easy recognition runs counter to our current ‘high performance’ culture, and it’s requisite demand for speedy assimilation of content, representing instead the poetic encounter with a potentiality or latency inherent in the nameless particularity of that which is.

Locally weighted learning model predictivek control for elastic joint robots

This paper proposes an efficient and online learning control system that uses the successful Model Predictive Control (MPC) method in a model based locally weighted learning framework. The new approach named Locally Weighted Learning Model Predictive Control (LWL-MPC) has been proposed as a solution to learn to control complex and nonlinear Elastic Joint Robots (EJR). Elastic Joint Robots are generally difficult to learn to control due to their elastic properties preventing standard model learning techniques from being used, such as learning computed torque control. This paper demonstrates the capability of LWL-MPC to perform online and incremental learning while controlling the joint positions of a real three Degree of Freedom (DoF) EJR. An experiment on a real EJR is presented and LWL-MPC is shown to successfully learn to control the system to follow two different figure of eight trajectories.

Harmonic elimination technique for a single-phase multilevel converter with unequal DC link voltage levels

Multilevel converters, because of the benefits they attract in generating high quality output voltage, are used in several applications. Various modulation and control techniques are introduced by several researchers to control the output voltage of the multilevel converters like space vector modulation and harmonic elimination (HE) methods. Multilevel converters may have a DC link with equal or unequal DC voltages. In this study a new HE technique based on the HE method is proposed for multilevel converters with unequal DC link voltage. The DC link voltage levels are considered as additional variables for the HE method and the voltage levels are defined based on the HE results. Increasing the number of voltage levels can reduce lower order harmonic content because of the fact that more variables are created. In comparison to previous methods, this new technique has a positive effect on the output voltage quality by reducing its total harmonic distortion, which must take into consideration for some applications such as uninterruptable power supply, motor drive systems and piezoelectric transducer excitation. In order to verify the proposed modulation technique, MATLAB simulations and experimental tests are carried out for a single-phase four-level diode-clamped converter.

Continuum mechanics modelling of microfilament networks with different architectures based on molecular investigation of single F-actin

The actin microfilament plays a critical role in many cellular processes including embryonic development, wound healing, immune response, and tissue development. It is commonly organized in the form of networks whose mechanical properties change with changes in their architecture due to cell evolution processes. This paper presents a new nonlinear continuum mechanics model of single filamentous actin (F-actin) that is based on nanoscale molecular simulations. Following this continuum model of the single F-actin, mechanical properties of differently architected lamellipodia are studied. The results provide insight that can contribute to the understanding of the cell edge motions of living cells.

Behavioral biometrics for persistent single sign-on

Driven by the rapid development of ubiquitous and pervasive computing, personalized services and applications are deployed to support our lives. Accordingly, the number of interfaces and devices (smartphone, tablet computer, etc.) provided to access and consume these services is growing continuously. To simplify the complexity of managing many accounts with different credentials, Single Sign-On (SSO) solutions have been introduced. However, a single password for many accounts represents a single-point-of-failure. Furthermore, once initiated SSO session is a high potential risk when the working station is left unlocked and unattended. In this paper, we present a conception of a Persistent Single Sign-On (PSSO) for ubiquitous home environments by involving the capabilities of Behavioral Biometrics to check the identity of the user continuously in an unobtrusive manner.

Intellectual property and joint ventures

This chapter outlines the most important ways in which intellectual property is protected in Australia, and also the factors which affect the rights of joint venture participants in the absence of specific agreement between such participants. It then examines particular issues which may be considered in preparing appropriate documentation for any joint venture which involves the utilisation or generation of intellectual property to ensure that the joint venture participants achieve their desired result in terms of the allocation of ownership and control of such rights. The analysis includes and explanation of the special considerations which affect co-operation in research between industry and a university or government research institution. Finally, the rights of the joint venturers to intellectual property upon termination of the joint ventures are considered.

UCPR r 360 – offers to settle – joint offer by first and second plaintiffs

The decision in ACN 070 037 599 Pty Ltd v Larvik Pty Ltd (No 2) [2008] QSC 118 involved a consideration of the implications for a plaintiff whose offer to settle under Part 5 of the Uniform Civil Procedure Rules 1999 (Qld) was made jointly with another plaintiff who abandoned her action before trial. The court found nothing wrong with the making of a joint offer. It concluded the successful plaintiff would be entitled to indemnity costs on the simple test of whether the judgment for that plaintiff was more favourable than the offer.

C-BN Single-Walled Nanotubes from Hybrid Connection of BN/C Nanoribbons: Prediction by ab initio Density Functional Calculations

We demonstrated for the first time by ab initio density functional calculation and molecular dynamics simulation that C0.5(BN)0.5 armchair single-walled nanotubes (NT) are gapless semiconductors and can be spontaneously formed via the hybrid connection of graphene/BN Nanoribbons (GNR/BNNR) at room temperature. The direct synthesis of armchair C0.5(BN)0.5 via the hybrid connection of GNR/BNNR is predicted to be both thermodynamically and dynamically stable. Such novel armchair C0.5(BN)0.5 NTs possess enhanced conductance as that observed in GNRs. Additionally, the zigzag C0.5(BN)0.5 SWNTs are narrow band gap semiconductors, which may have potential application for light emission. In light of recent experimental progress and the enhanced degree of control in the synthesis of GNRs and BNNR, our results highlight an interesting avenue for synthesizing a novel specific type of C0.5(BN)0.5 nanotube (gapless or narrow direct gap semiconductor), with potentially important applications in BNC-based nanodevices.

Formation of single-walled carbon nanotube via the interaction of graphene nanoribbons : ab initio density functional calculations

The interaction of bare graphene nanoribbons (GNRs) was investigated by ab initio density functional theory calculations with both the local density approximation (LDA) and the generalized gradient approximation (GGA). Remarkably, two bare 8-GNRs with zigzag-shaped edges are predicted to form an (8, 8) armchair single-wall carbon nanotube (SWCNT) without any obvious activation barrier. The formation of a (10, 0) zigzag SWCNT from two bare 10-GNRs with armchair-shaped edges has activation barriers of 0.23 and 0.61 eV for using the LDA and the revised PBE exchange correlation functional, respectively, Our results suggest a possible route to control the growth of specific types SWCNT via the interaction of GNRs.

An optimized online secondary path modeling method for single-channel feedback ANC systems

This paper proposes a new method for online secondary path modeling in feedback active noise control (ANC) systems. In practical cases, the secondary path is usually time varying. For these cases, online modeling of secondary path is required to ensure convergence of the system. In literature the secondary path estimation is usually performed offline, prior to online modeling, where in the proposed system there is no need for using offline estimation. The proposed method consists of two steps: a noise controller which is based on an FxLMS algorithm, and a variable step size (VSS) LMS algorithm which is used to adapt the modeling filter with the secondary path. In order to increase performance of the algorithm in a faster convergence and accurate performance, we stop the VSS-LMS algorithm at the optimum point. The results of computer simulation shown in this paper indicate effectiveness of the proposed method.

Nutritional analysis of gastric contents and body condition score at a single time point in feral horses in Australia

Objective: To determine the impact of a free-choice diet on nutritional intake and body condition of feral horses. Animals: Cadavers of 41 feral horses from 5 Australian locations. Procedures: Body condition score (BCS) was determined (scale of 1 to 9), and the stomach was removed from horses during postmortem examination. Stomach contents were analyzed for nutritional variables and macroelement and microelement concentrations. Data were compared among the locations and also compared with recommended daily intakes for horses. Results: Mean BCS varied by location; all horses were judged to be moderately thin. The BCS for males was 1 to 3 points higher than that of females. Amount of protein in the stomach contents varied from 4.3% to 14.9% and was significantly associated with BCS. Amounts of water-soluble carbohydrate and ethanol-soluble carbohydrate in stomach contents of feral horses from all 5 locations were higher than those expected for horses eating high-quality forage. Some macroelement and microelement concentrations were grossly excessive, whereas others were grossly deficient. There was no evidence of ill health among the horses. Conclusions and Clinical Relevance: Results suggested that the diet for several populations of feral horses in Australia appeared less than optimal. However, neither low BCS nor trace mineral deficiency appeared to affect survival of the horses. Additional studies on food sources in these regions, including analysis of water-soluble carbohydrate, ethanol-soluble carbohydrate, and mineral concentrations, are warranted to determine the provenance of such rich sources of nutrients. Determination of the optimal diet for horses may need revision.

Speaker diarization : "who spoke when"

Speaker diarization is the process of annotating an input audio with information that attributes temporal regions of the audio signal to their respective sources, which may include both speech and non-speech events. For speech regions, the diarization system also specifies the locations of speaker boundaries and assign relative speaker labels to each homogeneous segment of speech. In short, speaker diarization systems effectively answer the question of ‘who spoke when’. There are several important applications for speaker diarization technology, such as facilitating speaker indexing systems to allow users to directly access the relevant segments of interest within a given audio, and assisting with other downstream processes such as summarizing and parsing. When combined with automatic speech recognition (ASR) systems, the metadata extracted from a speaker diarization system can provide complementary information for ASR transcripts including the location of speaker turns and relative speaker segment labels, making the transcripts more readable. Speaker diarization output can also be used to localize the instances of specific speakers to pool data for model adaptation, which in turn boosts transcription accuracies. Speaker diarization therefore plays an important role as a preliminary step in automatic transcription of audio data. The aim of this work is to improve the usefulness and practicality of speaker diarization technology, through the reduction of diarization error rates. In particular, this research is focused on the segmentation and clustering stages within a diarization system. Although particular emphasis is placed on the broadcast news audio domain and systems developed throughout this work are also trained and tested on broadcast news data, the techniques proposed in this dissertation are also applicable to other domains including telephone conversations and meetings audio. Three main research themes were pursued: heuristic rules for speaker segmentation, modelling uncertainty in speaker model estimates, and modelling uncertainty in eigenvoice speaker modelling. The use of heuristic approaches for the speaker segmentation task was first investigated, with emphasis placed on minimizing missed boundary detections. A set of heuristic rules was proposed, to govern the detection and heuristic selection of candidate speaker segment boundaries. A second pass, using the same heuristic algorithm with a smaller window, was also proposed with the aim of improving detection of boundaries around short speaker segments. Compared to single threshold based methods, the proposed heuristic approach was shown to provide improved segmentation performance, leading to a reduction in the overall diarization error rate. Methods to model the uncertainty in speaker model estimates were developed, to address the difficulties associated with making segmentation and clustering decisions with limited data in the speaker segments. The Bayes factor, derived specifically for multivariate Gaussian speaker modelling, was introduced to account for the uncertainty of the speaker model estimates. The use of the Bayes factor also enabled the incorporation of prior information regarding the audio to aid segmentation and clustering decisions. The idea of modelling uncertainty in speaker model estimates was also extended to the eigenvoice speaker modelling framework for the speaker clustering task. Building on the application of Bayesian approaches to the speaker diarization problem, the proposed approach takes into account the uncertainty associated with the explicit estimation of the speaker factors. The proposed decision criteria, based on Bayesian theory, was shown to generally outperform their non- Bayesian counterparts.