Recent work on the use of discrete first-order approximate models of plant dynamics in multi-vari... more Recent work on the use of discrete first-order approximate models of plant dynamics in multi-variable process control is extended to consider the effect on transient performance non-linearities such as measurement quantization or deadzone at the implement stage. Explicit upper bands on the transient error induced by the nonlinearity are obtained and their validity illustrated by nemerical examples. In particular, at fast sampling rates, it is seen that, not only does the first-order controller produce a closed-loop system with rapid, non-oscillating responses and small interaction effects, it also ensures that peak transient errors due to the non-linearity are less than the error involved in approximating the non-linearity by a unity grain matrix. For example, if the nonlinearity is measurement quantization, the peak transient effects of the quantization will be bounded by the quantization error itself at fast sampling rates, and will not be amplified by system dynamics.
Resonance Audio is an open source project designed for creating and controlling dynamic spatial s... more Resonance Audio is an open source project designed for creating and controlling dynamic spatial sound in Virtual & Augmented Reality (VR/AR), gaming or video experiences. It also provides integrations with popular game development platforms and digital audio workstations (as a preview plugin). Resonance Audio binaural decoder is used in YouTube to provide binaural rendering of 360/VR videos. This paper describes the core sound spatialization algorithms used in Resonance Audio and can be treated as a companion to the Resonance Audio C++ / MATLAB library source code.
IET Irish Signals and Systems Conference (ISSC 2006), 2006
Abstract Accurate localization of objects in an area has many important applications such as spea... more Abstract Accurate localization of objects in an area has many important applications such as speaker localization and in self-reconfigurable networks systems. Methods using cross-correlation at audio and ultrasonic frequencies are prone to false or inaccurate results for ...
Recent studies have shown the effectiveness of using Machine Learning (ML) to perform Sound Sourc... more Recent studies have shown the effectiveness of using Machine Learning (ML) to perform Sound Source Localization (SSL) tasks. However, most of this research has been primarily aimed at performing SSL in the azimuthal plane only. While the Interaural Cues which allow us to perceive azimuthal source location are well researched, cues which allow us to determine elevation location are understood to a lesser degree. It is generally regarded that spectral cues resulting from reflections of sound from the head, shoulders and pinnae are responsible for our perception of elevated sound. In this study we use Machine Learning to determine the proficiency of a variety of Interaural and Spectral cues in determining sound source elevation.
This paper further investigates previous findings that coefficients of acoustic responses can be ... more This paper further investigates previous findings that coefficients of acoustic responses can be modelled as random polynomials with certain constraints applied. In the case of room impulse responses, the median value of their clustered roots has been shown to be directly related to the reverberation time of the room. In this paper we examine the frequency dependency of reverberation time and we also demonstrate the method’s robustness to truncation of impulse responses.
Recent work on the use of discrete first-order approximate models of plant dynamics in multi-vari... more Recent work on the use of discrete first-order approximate models of plant dynamics in multi-variable process control is extended to consider the effect on transient performance non-linearities such as measurement quantization or deadzone at the implement stage. Explicit upper bands on the transient error induced by the nonlinearity are obtained and their validity illustrated by nemerical examples. In particular, at fast sampling rates, it is seen that, not only does the first-order controller produce a closed-loop system with rapid, non-oscillating responses and small interaction effects, it also ensures that peak transient errors due to the non-linearity are less than the error involved in approximating the non-linearity by a unity grain matrix. For example, if the nonlinearity is measurement quantization, the peak transient effects of the quantization will be bounded by the quantization error itself at fast sampling rates, and will not be amplified by system dynamics.
Resonance Audio is an open source project designed for creating and controlling dynamic spatial s... more Resonance Audio is an open source project designed for creating and controlling dynamic spatial sound in Virtual & Augmented Reality (VR/AR), gaming or video experiences. It also provides integrations with popular game development platforms and digital audio workstations (as a preview plugin). Resonance Audio binaural decoder is used in YouTube to provide binaural rendering of 360/VR videos. This paper describes the core sound spatialization algorithms used in Resonance Audio and can be treated as a companion to the Resonance Audio C++ / MATLAB library source code.
IET Irish Signals and Systems Conference (ISSC 2006), 2006
Abstract Accurate localization of objects in an area has many important applications such as spea... more Abstract Accurate localization of objects in an area has many important applications such as speaker localization and in self-reconfigurable networks systems. Methods using cross-correlation at audio and ultrasonic frequencies are prone to false or inaccurate results for ...
Recent studies have shown the effectiveness of using Machine Learning (ML) to perform Sound Sourc... more Recent studies have shown the effectiveness of using Machine Learning (ML) to perform Sound Source Localization (SSL) tasks. However, most of this research has been primarily aimed at performing SSL in the azimuthal plane only. While the Interaural Cues which allow us to perceive azimuthal source location are well researched, cues which allow us to determine elevation location are understood to a lesser degree. It is generally regarded that spectral cues resulting from reflections of sound from the head, shoulders and pinnae are responsible for our perception of elevated sound. In this study we use Machine Learning to determine the proficiency of a variety of Interaural and Spectral cues in determining sound source elevation.
This paper further investigates previous findings that coefficients of acoustic responses can be ... more This paper further investigates previous findings that coefficients of acoustic responses can be modelled as random polynomials with certain constraints applied. In the case of room impulse responses, the median value of their clustered roots has been shown to be directly related to the reverberation time of the room. In this paper we examine the frequency dependency of reverberation time and we also demonstrate the method’s robustness to truncation of impulse responses.
Uploads