Last Updated: 20060822, mei
P22 - Posters: Signal Processing
Sunday, October 8, 11:30 am — 1:00 pm
P22-1 Decoding Second Order Ambisonics to 5.1 Surround Systems—Martin Neukom, Zurich School of Music, Drama and Dance, HMT - Zurich, Switzerland
In order to play back Higher Order Ambisonics (HOA) in concert, symmetric loudspeaker set-ups with a large number of speakers are used. At the moment the only possibilities to provide Ambisonics to home users are the rendering for headphones with HRTF and the conversion to 5.1 surround systems. This paper shows the difficulties and limitations of the conversion of Higher Order Ambisonics to 5.1 surround and presents some viable solutions.
Convention Paper 6980 (Purchase now)
P22-2 Artificial Reverberation: Comparing Algorithms by Using Binaural Analysis Tools—Joerg Bitzer, Denis Extra, University of Applied Science Oldenburg - Oldenburg, Germany
Different measures are known for objective measurements of the spatial quality of concert halls . Many of these measures are based on analyzing the binaural impulse responses. In this paper we will compare different algorithms for artificial reverberation in terms of these measures. The tested algorithms are commercially available devices and digital plug-ins in a broad price range. For the analysis, we programmed an analysis toolbox that contains several binaural analysis methods, including the interaural cross-correlation and the interaural difference. Furthermore, lesser known measures, modifications, and new techniques will be presented. The results indicate that objective measures can give some first impression of the spatial quality of reverberation devices.
Convention Paper 6981 (Purchase now)
P22-3 Loudspeaker and Room Response Modeling with Psychoacoustic Warping, Linear Prediction, and Parametric Filters—Sunil Bharitkar, Chris Kyriakakis, Audyssey Laboratories, Inc. - Los Angeles, CA, USA
Traditionally, room response modeling is performed to obtain lower order room impulse response models for real-time applications. These models can be FIR or IIR, and may be either linear-phase or minimum-phase. In this paper we present an approach to model room responses using linear predictive coding (LPC) and parametric filters designed in the frequency warped domain. Frequency warping to the psychoacoustic Bark scale allows significant lower filter order designs. Within this context, the LPC model utilizes a significantly lower number of poles to model room resonances at low frequencies in the warped domain. The relatively low-order LPC pole locations and gains are then used to determine the center frequencies, the gain, and Q of a parametric filter bank. Gain and Q optimization of the parametric filter bank is performed to match the parametric filter spectrum to the LPC spectrum. Subsequently, the second-order poles and zeros of the parametric filter bank are directly unwarped back into the linear domain for low-complexity real-time applications. The results show that warping lowers the computational requirements for determining the roots as the density of the roots, and the number of roots of the LPC polynomial is substantially reduced. Furthermore, results fro sing simply four-to-six parametric filter banks, modeled from the LPC spectrum, below 400 Hz show significant equalization.
Convention Paper 6982 (Purchase now)
P22-4 Contactless Hearing Aid for Infants Employing Signal Processing Algorithms—Maciej Kulesza, Piotr Dalka, Gdansk University of Technology - Gdansk, Poland; Bozena Kostek, Gdansk University of Technology - and Institute of Physiology and Pathology of Hearing, Gdansk, Poland
The proposed contactless hearing aid is designed to be attached to the infant’s crib for sound amplification in a free field. It consists of a four-electret microphone matrix and a prototype DSP board. The compressed speech is transmitted and amplified via miniature loudspeakers. Algorithms that are worked out deal with parasitic feedback, which occurs due to the small distance between microphone and monitors and potentially high amplification required. The beamforming algorithm is based on an artificial neural network (ANN). The ANN is used as a nonlinear filter in the frequency domain. Principles of algorithms engineered and the prototype DSP unit design are presented in the paper. Also, results of experiments simulating the real-life conditions are analyzed and discussed.
Convention Paper 6983 (Purchase now)
P22-5 An Enhanced Implementation of the ADRess (Azimuth Discrimination and Resynthesis) Music Source Separation Algorithm—Rory Cooney, Niall Cahill, Robert Lawlor, National University of Ireland - Maynooth, Co. Kildare, Ireland
In this paper we present a novel enhancement to an existing music source separation algorithm that allows for a 76 percent decrease in computational load while enhancing its separation capabilities. The enhanced implementation is based on the ADRess (Azimuth Discrimination and Resynthesis) algorithm, which performs a separation of sources within stereo music recordings based on the spatial audio cues created by source localization techniques. The ADRess algorithm employs gain scaling and phase cancellation techniques to isolate sources based on their position across the stereo field. Objective measures and subjective listening tests have shown the separation performance of the enhanced algorithm to be objectively and perceptually comparable with that of the original ADRess algorithm, while realizing a finer spatial resolution.
Convention Paper 6984 (Purchase now)
P22-6 A Simple, Robust Measure of Reverberation Echo Density—Jonathan Abel, Universal Audio, Inc. - Santa Cruz, CA, USA, Stanford, University, Stanford, CA, USA; Patty Huang, Stanford University - Stanford, CA, USA, Helsinki University of Technology, Espoo, Finland
A simple, robust method for measuring echo density from a reverberation impulse response is presented. Based on the property that a reverberant field takes on a Gaussian distribution once an acoustic space is fully mixed, the measure counts samples lying outside a standard deviation in a given impulse response window and normalizes by that expected for Gaussian noise. The measure is insensitive to equalization and reverberation time and is seen to perform well on both artificial reverberation and measurements of room impulse responses. Listening tests indicate a correlation between echo density measured in this way and perceived temporal quality or texture of the reverberation.
Convention Paper 6985 (Purchase now)