Impact case study (REF3b) Page 1 Institution: University of Derby Unit of Assessment: UoA 15 Title of case study: Improved Spatial Audio from Ambisonic Surround Sound Software 1. Summary of the impact (indicative maximum 100 words) The reduction of spatial variation in the quality of reproduced sound within a defined space using varied loudspeaker placements is a significant challenge for sound engineers. Dr Bruce Wiggins has conducted research into encoding, decoding and processing algorithms using Ambisonics, a system based around full-sphere sound reproduction. The outcomes of the research have been made accessible to the wider community by the creation of a suite of software plug-ins (WigWare), a production workflow, and associated teaching materials which can enable commercial audio workstations to benefit from Ambisonics. There are numerous recorded instances of successful use. 2. Underpinning research (indicative maximum 500 words) Ambisonics, pioneered in the 1980’s by Michael Gerzon, is a kernel-based 3D surround sound system. The encoding (recording or panning) of the audio is separated from the decoding (or rendering) of the audio to speaker feeds. This means that the system can be rendered to any number of speakers in almost any position in 3D space, as long as the positions of the speakers are known. Moreover, Ambisonics is a system optimised around a number of psycho-acoustic criteria which, when implemented, reduce the variability of audio no matter what speaker arrangement is used for reproduction. This allows for a ‘mix once’ system where subsequent remixing is not necessary when replayed over different loudspeaker systems. The research work carried out began by giving a thorough treatise on Ambisonic theory, specifying the relationship and conversions from and to other surround systems, which led to work quantifying and optimising the system using measured Head Related Transfer Function (HRTF) data and the more standard energy and velocity vector analysis. These complex multiple input, multiple output system were then optimised using a modified heuristic Tabu search algorithm with multiple defined fitness functions (Wiggins 2003, 2004). This was necessary as, up to this point (2003), there was no known or published method of generating Ambisonic decoder coefficients for irregular speakers arrays. This included the standard 5 speaker array as specified by the International Telecommunication Union (ITU-R BS 775); the standard matrix pseudo-inverse approach gave results that were psycho-acoustically sub-optimal as the matrices in question became ill- conditioned. Work carried out by Gerzon just before his death in 1995 on this subject was shown to lead to sub-optimal results (Wiggins 2003). This research solved that problem. Subsequent work concentrated on the encoding side of Ambisonics, giving new insight in the performance of Ambisonic, 3D microphones with respect to near-field effects (distance cues), and how these could be handled in terms of decoding to loudspeaker arrays (Wiggins, 2009). Concurrently, work based on both the use of Ambisonic mixing tools in the context of music production and digital audio workstations allowed for the wider community to utilise the outputs of this research in the form of software Virtual Studio Technology (VST) plug-ins which implemented encoding, decoding and 3D reverb effects for up to 24 loudspeakers for 1st, 2nd, 3rd and 4th order Ambisonic systems. Note here that the order is an indication of the number of channels used in the encoding [(N+1)^2 where N is the order], with higher channel counts resulting in better spatial acuity. This software, named WigWare, first released in November 2005, has evolved to take in further work based on proximity effect, near-field compensation and speaker compensation. This arises from the distance of both the encoded, or recorded source, and the speakers used to replay the audio. To disseminate the work, ‘screencasts’ documenting how to set up and use the software research outputs in an example Digital Audio Workstation (DAW) have been made publicly available (using Reaper as the DAW – http://reaper.fm). This work has led to a number of invited presentations, at DAFx 2012, Institute of Acoustics Reproduced Sound 2012 and Birmingham University and involvement with the BBC R&D Audio Research Partnership. The principle investigator was Bruce Wiggins who enrolled as a PhD student at the University of Derby in 1999, completing in 2004. The supervisors were Iain Paterson-Stephens and Professor Richard Thorn with contributing work by Dr Stuart Berry and Dr Valerie Lowndes (all staff at the University of Derby)