EASAIER - Enabling Access to Sound Archives through Integration, Enrichment and Retrieval
Many digital sound archives suffer from tremendous problems concerning access. Sound materials are often held separately from other materials and media, they cannot easily be listened to or browsed, and there is no way to search the content. Existing systems which attempt to deal with these issues are often library or content specific, of limited functionality, or difficult to use.
Extending over two and a half years, the EASAIER project will enable enhanced access to sound archives by providing multiple methods of retrieval, integration with other media archives, content enrichment and enhanced access tools. It offers methods of searching content based on audio features, musical features, or speech content. EASAIER also supports cross-media retrieval, enabling access to other media besides just audio. It implements recent advances in machine learning, music and speech processing, and information retrieval. Furthermore, it addresses a growing demand for interactive electronic materials. www.easaier.org

VOCAL - Voice Origin Creation for Animated Life
Animated movie and computer gaming are growing, competitive industries with a consequent need to improve productivity and enrich the experience of users in order to retain old and capture new markets. The objective of the VOCAL project team is to develop a Digital Signal Processing (DSP) speech synthesiser which will provide a flexible means for generating, altering and transforming speech based on physical attributes such as perceived size, gender, age and emotion of an animated character. With this speech synthesiser animators may generate a voice origin, appropriate to the physical traits of a character, as required, without the time or financial costs associated with hiring a professional actor. Furthermore, a character's voice could adapt to its ever changing virtual environment with the emotional content being modified accordingly. www.audioresearchgroup.com/vocal

Intelligent Audio Environments
The intelligent audio environments project focuses on personal profiling for automatic playlist generation. Many people currently have personal digitised music collections amounting to several hundreds or even thousands of hours. Browsing such collections is becoming ever more tedious. This project aims to profile users listening habits based on current environmental and ambient data such as time, day, month, season, weather conditions, lighting conditions, location and activity. Much of this data can be obtained dynamically online and is updated regularly. The remaining local data is to be captured by a system of sensors. We hypothesise that these environmental conditions have a significant effect on mood and that music can be chosen to reflect or oppose this mood. Each time the user makes a selection, the ambient and environmental data is captured and stored as metadata for the chosen track. In conjunction with this, music similarity algorithms are used to generate clusters of songs which have musicologically similar features and thus theoretically can be represented by the same ambient/environmental metadata. Automatic playlist generation is then achieved by selecting song clusters from the user's collection which exhibit similar ambient metadata to that which is currently present in the users listening environment. This method of correlating environmental data with musical metadata is aimed at overcoming the problems associated with attaching explicit emotion descriptions to specific types or pieces of music.

Adaptive Music Techniques for Computer Games
The non-linear nature of video games poses interesting and challenging problems for game music composers and developers. While the composition of a linear score follows a path known in advance, a score for a video game should ideally take into account the unpredictability of the player's actions and also the effect those actions have on the game play world. While the scoring of themes and cues for specific game play events and locations can follow the traditional linear composition processes, the main challenge for game music composers and programmers is to provide a soundtrack that will seamlessly reflect and possibly predict the state of the player and game world. Despite the fact that the notion of adaptive music has been in the game music industry for over ten years, it is only recently that a formal definition has been regarded as a necessity in order to help identify the problem space. The challenge of developing a definition has been compounded by the widespread use of terms similar but different to adaptive music such as interactive music and dynamic music. This 3.5 year internally funded project aims to explore, formalise and define a software based system which provides an intuitive way to compose adaptive music for the purposes of computer gaming.

IMAAS - Interactive Music Archive Access System
The objective of this project is to develop an on-line system which allows remote interactive access to audio and music archives. The system will provide value-added services such as improved search functionality and increased metadata availability. The system will also feature state of the art digital signal processing technology which will allow the user to interact with the archive by processing, manipulation and visualisation of the audio content. Initially, the system will be focused on Irish traditional music. The Irish Traditional Music Archive will be used as a testbed for the completed system. www.audioresearchgroup.com/imaas

DiTME - Digital Tools For Music Education
The principle innovative aspect of the DITME project is the development of a powerful, yet user-friendly PC-based toolkit for music teaching and learning.
While very few trainee musicians are able to play a difficult piece at 'full' speed, many can play at 80 - 90% of full speed. One way to help the less advanced student would be to enable the recording to be slowed to any desired tempo without changing the key. Such a facility requires independent speed and pitch control and is the first tool of the system.
The second tool in the system is a facility to allow a lead instrument to be separated out of an ensemble recording. This is to enable any ensemble recording to be transformed into an MMO format.
The third and final tool in the system is a music transcription facility. This tool works in conjunction with the lead instrument separation tool and converts the separated lead into music score.