 |
  |
  |
  |
  |
  |
  |
  |
 |
  Publications |  | |
|
| Details of publications to date: |
|
Full versions of some papers are available in PDF format. To view these, you will need Adobe Acrobat Reader. In reverse chronological order:
2002 Audio short course notes. Published 11th February 2002.
Replay Gain - A Proposed Standard. Published on line 10th July 2001. Abstract: For 20 years, the movie industry has created sound tracks designed to be replayed at a fixed, calibrated monitor gain. The audio industry has no such luxury, and as a result the loudness of a given audio track has more to do with the year of issue or the whim of the producer than the intended emotional effect. The Replay Gain proposal contains simple methods of calculating and representing the ideal replay gain for every track and album.
Psychoacoustic Models and Non-linear Human Hearing. Presented at the 109th Conference of the Audio Engineering Society, Los Angeles, September 2000. Abstract: Non-linearity in the human ear can cause audible distortion not present in the original signal. Such distortion is generated within the ear by inter-modulation of a spectral complex, itself containing possible masked components. When psychoacoustic codecs remove these supposedly masked components, the in-ear-generated distortion is also removed, and so our listening experience is modified. In this paper, the in-ear distortion is quantified and a method suggested for predicting the in-ear distortion arising from an audio signal. The potential performance gains due to incorporating this knowledge into an audio codec are assessed.
MPEG-1 layer 3 decoder accuracy tests. Published on line 14th July 2000. Abstract: It is well known that mp3 encoders can have a significant effect on the sound quality of coded audio. However, the MPEG standard sets out the requirement for a decoder exactly, so little attention has been given to the quality and accuracy of mp3 decoders. On this website, we test a number of common software mp3 decoders and players, and show that the majority fail to correctly decode MPEG-1 layer 3 audio files.
Time-domain auditory model for the assessment of high-quality coded audio. Presented at the EPSRC sponsored conference Prep 2000, Nottingham, April 2000. Abstract: An auditory model for assessing the perceived quality of coded audio signals is described. This model simulates the functionality of the human ear, as well as its characteristics, to yield an accurate prediction of human perception. This paper won the Philips sponsored prize for Best paper in Electronics. It is a condensed version of the paper of the same name given at the 107th AES conference.
The Human Auditory System. Lecture notes to accompany the Network Services short course, University of Essex, March 2000. Abstract: In this lecture, we will examine the apparatus used by a human to hear sound. These notes include the diagrams used within the lecture, as well as a description of the material covered. Further reading is suggested at the end. You will not be examined on the contents of this lecture, but knowledge of the human auditory system will aid you in your understanding of the various audio-coding schemes that will be covered tomorrow.
Time-domain auditory model for the assessment of high-quality coded audio. Presented at the 107th Conference of the Audio Engineering Society, New York, September 1999. Abstract: A time-domain auditory model is described. The spectral analysis function of the inner ear is simulated by a non-linear filter bank, while the temporal response is simulated by complex filtering of the envelope. The spectral and temporal masking properties of the model are calibrated with known data. The effectiveness of this model in predicting the perceived quality of coded audio streams is examined.
A Binaural simulation which renders out of head localisation with low cost digital signal processing of Head Related Transfer Functions and pseudo reverberation. Presented at the 104th Conference of the Audio Engineering Society, Amsterdam, May 1998. Abstract: A binaural simulation is developed on a single 40MHz D.S.P. chip, which renders out of head localisation, by filtering the signal with a pair of diffuse normalised Head Related Transfer Functions, and adding slight reverberation. An informal subjective assessment revealed the actual localisation varied between listeners, front/back confusion and side domination being two common problems. However, externalisation is generally achieved. |
