With a budget of over 600K Euros and a pre-financing of 395K Euros, the project involves a partnership with the Digital Media Centre and the Audio Research Group at DIT, one of Europeís leading audio Codec developers (DLLNI Ltd based in Bangor, Northern Ireland) and a specialist SME audio post-production company (Tamborine Productions Ltd based in London). Awarded in November 2009 through the Seventh Framework Programme of the European Union through the Support for Training and Career Development of Researchers (Marie Curie) and the Industry-Academia Partnerships and Pathways initiative, the project is coordinated and directed by Principal Investigators Dr Charlie Cullen (Digital Media Centre) and SFI Stokes Professor Jonathan Blackledge (Information and Communications Security Research Group), respectively. The broad aim of the research project is to use object-oriented audio to overcome many of the problems associated with channel-based coding and make it possible to transmit 3D audio scenes and objects over narrow bandwidths in a way that allows real user interactivity.
Research will focus on applications of Auditory Scene Analysis, a process whereby the human auditory system organizes sound into perceptually meaningful elements. This has the potential to drive much simpler and more bandwidth-efficient coding methods, based on human cognitive models. Another goal of the four year project is to train researchers in the development of algorithms, Codecs and tools for the 3D audio technology required to deliver very high-quality, object driven sound for immersive and interactive entertainment. Further, through Tamborine Productions Limited, the ultimate goal is to commercialize the technologies developed for the rapidly expanding markets in the digital media sector.
Auditory scene analysis (ASA) studies the interactions of concurrently presented sounds and their perception by the listener. The human listener splits and fuses physical sounds into logical groups known as streams, where multiple streams containing similar sounds (such as a choir) can often be perceived as a single stream. ASA describes stream grouping using psychological terms such as familiarity (a well known song is easier to recognize), similarity (a group of singing voices) and proximity (sources that are close in pitch, intensity or location).
These terms (and others) are often related to acoustic parameters such as pitch, intensity and timbre and help to define variations in an auditory scene over time. An example of this is given in the figure below which shows possible groups in grey within Mozartís Requiem using ASA. The score is of the first four bars of the Introitus Requiem Aeternam for contrabass, strings and wind instruments.
In this example, an initial rhythmic juxtaposition of bass and strings defines two streams in bar one. At the same time, the interplay between these streams also groups them at another level relative to the bassoon melody. The introduction of a violin part with the same rhythm as the other stringed instruments creates a single string group in bar two. At the same time, the rhythm and pitch contour of the initial bassoon melody is repeated in the second bar by the basset horns.
Although this provides a cue that is familiar and serves to group the bassoon and basset horns together, it also provides a harmonic relation to the bassoon part that separates them. Mozart provides two melodic voices that mimic and harmonize over an accompanying section, defining two overall streams that subdivide into more complex interactions at instrument level.
Most audio coding algorithms adopt a channel-oriented approach (e.g. MPEG-1 and MPEG-2), where the final mixed signal in a given number of channels is processed for transmission. Channel coding is often lossy and destructive. For example, the relative volume balance between contrabass, strings and woodwind stream objects in Mozartís Requiem will be defined by an engineer during mixing. If changes are made to the playout setup, or bandwidth is reduced to stream to a mobile device, the priorities in the original are lost in the adaptation to the new transmission conditions. Lossy compression is likely to have a worse effect on the woodwind than the strings or contrabass, and will severely compromise the final content.
Algorithms such as MPEG-4 and AudioBIFS can transmit a 3D scene comprised of audio objects to a wide range of playout setups, offering interactivity and capable of low bandwidth operation. AudioBIFS can mix, process and spatialise audio streams, while advanced AudioBIFS (AABIFS) also support descriptions of the acoustic properties of sound sources and environments, the source direction and room reverberation.
However, the MPEG-4 standard is extremely complex, very few authoring tools exist and attempts to implement an MPEG-4 AABIFS player remain experimental. In addition, the prioritisation and interaction of sources within an MPEG-4 object are not clearly defined from either a source or channel coding perspective. In consequence, it is entirely feasible that a priority source (e.g. main character dialogue) could noticeably drop in quality under limited bandwidth conditions in order to preserve other sources (e.g. crowd chatter) that are relatively unimportant to the real auditory scene. The research project proposes to advance the state of the art by using ASA parameters, defined during perception testing and to build much simpler algorithms for the creation of 3D audio objects. Common descriptors from the existing MPEG-4 AABIFS set will be resolved with these ASA parameters to produce a straightforward model for the 3D audio object that is compatible with existing standards.
Security economics is the analysis of aggregate risks facing society and economy using rigorous analytical and empirical tools of economics. Policy makers may tend to take imperfect security decisions (e.g. regulations) based on a public perception of (in)security, with an impact to market structures. A singular focus on security or competitiveness would be too narrow and research under this area offers key insights that will contribute to balancing security and the overall policy objectives. Economic theory in particular can offer key insights, enabling governments to optimise their efforts to enhance security and growth. The expected impact of undertaking research in Security Economics is the provision to users of a decision support system providing them for insight into the pros and cons of specific security investments compared to a set of alternatives taking into account a wider societal context. This requires research projects that involve the exhaustive analysis of time series analysis that are best associated securing accurate economic forecasting.
The development of financial models that can provide increasingly accurate forecasts has been a fundamental requirement in economics for many years. For example, macroeconomic models are used in the preparation of Budget and Pre-Budget Report forecasts by most governments. Given the increased volatility in the financial markets, any approach that can provide more accurate forecasts are of interest to those working in the financial sector. Current models for market analysis are based on the Efficient Market Hypothesis EMH, which involves assumptions that are incompatible with real financial signals, including that market signals exhibit stationary Gaussian processes. The EMH is the basis for models such as the Black-Scholes equation.
The research question considered which underpins this project is whether a Fractal Market Hypothesis (FMH), based on the application of Fractional Dynamics can provide a more accurate representation of financial signals, in genera. Fractional Dynamics relies on models that include fractional Partial Differential Equations. This proposal is based on investigating solutions to a non-stationary fractional diffusion equation whose characteristics may be used to model financial signals with greater accuracy and physical significance than current models do. This approach may provide the basis for developing new macroeconomic forecasting measures that are both accurate and robust.
Predicting the markets has become one of the most important problems in financial engineering. While quasi-deterministic models can be of value in understanding micro-economic systems (with known Ďoperational conditions'), in a global economy, we can take advantage of the scale of the Ďsystemí to describe its behaviour in terms of functions of random variables. Further, if the statistical characteristics of the data are scale invariant (as is the case with macroeconomic signals), then we can make use of fractal models to interpret the data. This is the basis for the FMH which is the underlying theme associated with this proposal in which the principal questions to be addressed are:
Following the sale of an exclusive license by Dublin Institute of Technology to Currency Traders Ireland Limited (www.tradersnow.com), work has began on integrated a new and unique set of indicators into MetaTrader 4, a financial analysis package that provides real time on-line access to all major currency exchange rates and is used world wide. The indicators compliment those that have already been implement by Currency Traders Ireland Limited, an SME with many years experience in FOREX trading and, in addition to the companies growing portfolio of market indicators, provides on-line training to a world-wide network of currency traders.
The GUI of MetaTrader 4 with the new buy/sell indicatorís developed for Currency Traders Ireland Limited.