Last Updated: 20060821, mei
P13 - Signal Processing - Part 1
Saturday, October 7, 8:30 am — 12:30 pm
Chair: Ronald Aarts, Philips Research Laboratories - Eindhoven, The Netherlands
P13-1 Picturing Dither: Dithering Pictures—Stanley Lipshitz, Cameron Christou, University of Waterloo - Waterloo, Ontario, Canada
The desirable properties that follow from the use of (nonsubtractive) triangular probability density function (TPDF) random dither in digital audio quantization and noise shaping are now well known in the audio community. The principal purpose of this paper is to use a visual analogy to aid audio engineers in their understanding of how proper TPDF dithering and noise shaping can convert otherwise objectionable, correlated quantization errors into benign, uncorrelated, and less visible ones. As they say, “a picture is worth a thousand words.” Our secondary purpose is to demonstrate, in the process, that the very same concepts, applied now in the spatial instead of the temporal domain, are just as useful and beneficial in the field of digital picture processing too. We present color and monochrome images of the results of coarse quantization, both with and without dither and/or noise shaping, to help us make our points. [In the “live” presentation of this paper, we shall play an audio example at the same time as we show each picture, so that one can simultaneously both see and hear each effect being discussed.]
Convention Paper 6923 (Purchase now)
P13-2 Comparison of Frequency-Warped Representations for Source Separation of Stereo Mixtures—Juan José Burred, Thomas Sikora, Technical University Berlin - Berlin, Germany
We evaluate the use of different frequency-warped, nonuniform time-frequency representations for the purpose of blind sound source separation from stereo mixtures. Such transformations enhance resolution in spectral areas relevant for the discrimination of the different sources, improving sparsity and mixture disjointness. In this paper we study the effect of using such representations on the localization and detection of the sources, as well as on the quality of the separated signals. Specifically, we evaluate a constant-Q and several auditory warpings in combination with a shortest path separation algorithm and show that they improve detection and separation quality in comparison to using the Short Time Fourier Transform.
Convention Paper 6924 (Purchase now)
P13-3 Auditory Component Analysis—Jon Boley, University of Miami - Coral Gables, FL, USA
Two of the major research areas currently being evaluated for the so-called sound source separation problem are auditory scene analysis and a class of statistical analysis techniques known as independent component analysis. This paper presents a methodology for combining these two techniques. It suggests a framework that first separates sounds by analyzing the incoming audio for patterns and synthesizing or filtering them accordingly It then measures features of the resulting tracks and separates the sounds statistically by matching feature sets and attempting to make the output streams statistically independent. The proposed system is found to successfully separate artificial and acoustic mixes of sounds. As expected, the amount of separation is inversely proportional to the amount of reverberation present, number of sources, and interchannel correlation.
Convention Paper 6925 (Purchase now)
P13-4 Frequency Domain Artificial Reverberation Using Spectral Magnitude Decay—Earl Vickers, The Sound Guy, Inc. - Seaside, CA, USA; Jian-Lung (Larry) Wu, Stanford Center for Computer Research in Music and Acoustics - Stanford, CA, USA; Praveen Gobichettipalayam Krishnan, Ravirala Narayana Karthik Sadanandam, University of Missouri - Rolla, MO, USA
A novel method of producing artificial reverberation in the frequency domain, using spectral magnitude decay, is presented. The method involves accumulating the magnitudes of the short-time Fourier transform, based on the desired decay time as a function of frequency. Compared to time domain methods such as feedback delay networks, the current method requires less memory and provides independent control of the reverberation energy and decay time in each frequency bin. Compared to convolution reverbs, the current approach offers flexible parametric control over the decay spectra and a computational cost that is independent of decay time.
Convention Paper 6926 (Purchase now)
P13-5 Design of an Automatic Beat-Matching Algorithm for Portable Media Devices—Danny Jochelson, Texas Instruments, Inc. - Dallas, TX, USA; Stephen Fedigan, General Dynamics Vertex RSI - Richardson, TX, USA
Methods to achieve accurate beat detection for musical signals have received much attention recently; however, very little literature has addressed techniques for achieving beat matching between two streams on portable devices with limited memory and processing power. This paper describes the architecture, design methods, obstacles, optimizations, and results for a new beat matching algorithm created for real-time use on embedded devices. This algorithm produces promising performance for use on portable media devices that often play modern musical genres.
Convention Paper 6927 (Purchase now)
P13-6 Artificial Reverberation: Comparing Algorithms by Using Monaural Analysis Tools—Denis Extra, Uwe Simmer, Sven Fischer, Joerg Bitzer, University of Applied Science Oldenburg/Ostfriesland/Wilhelmshaven - Oldenburg, Germany
In this paper a comparison of different algorithms for artificial reverberation is presented. The tested algorithms are commercially available devices and digital plug-ins in a broad price range plus algorithms known from literature. For the analysis we developed an analysis toolbox, which contains several monaural analysis methods, including the energy decay curve in fractional octave bands, auto-correlation, and other known measures of reverberation qualities. Furthermore, the behavior over time will be analyzed, showing that many systems cannot be considered as time-invariant. Some statistical analysis of the impulse response will also be given. The purpose is to investigate whether synthetic reverberation is created pertaining to attributes of real rooms and whether there are differences between algorithms or not.
Convention Paper 6928 (Purchase now)
P13-7 Inverse Filtering Design Using a Minimal-Phase Target Function from Regularization—Scott Norcross, Communications Research Centre - Ottawa, Ontario, Canada; Martin Bouchard, University of Ottawa - Ottawa, Ontario, Canada; Gilbert Soulodre, Communications Research Centre - Ottawa, Ontario, Canada
Inverse filtering methods commonly use amplitude regularization as a technique to limit the amount of work done by the inverse filter. The amount of regularization needed must be carefully selected so that the audio quality is not degraded. This paper introduces a method of using the magnitude of the regularization to design a target/desired response in which the phase response can be arbitrarily chosen. By choosing a minimum-phase response, one can reduce any pre-response in the corrected signal that is introduced by the regularization. A phase response that consists of a frequency-dependent mixture of minimum- and zero-phase components is also introduced. Informal listening tests were performed to verify the effectiveness of the new method.
Convention Paper 6929 (Purchase now)
P13-8 The Origins of DSP and Compression: Some Pale Gleams from the Past—Jon Paul, Scientific Conversion, Inc. - Novato, CA, USA
This paper explores the history that led to modern day DSP and audio compression. The roots of of modern digital audio sprang from Dudley’s 1936 VOCODER and the WWII-era SIGSALY speech scrambler. We highlight these key inventions, detail their hardware and block diagrams, describe how they functioned, and illustrate their relationship to modern day DSP and compression algorithms.
Convention Paper 6930 (Purchase now)