Last Updated: 20050419, meiSaturday, May 28, 09:30 — 11:00
Z1-1 A Listening Test of Dither in Audio Systems—Preben Kvist, DELTA Acoustics & Vibration - Hørsholm, Denmark; Karsten Bo Rasmussen, Oticon A/S - Hellerup, Denmark; Torben Poulsen, Technical University of Denmark - Lyngby, Denmark
This paper investigates the subjective effects of the use of dither in digital audio systems. A short introduction to dithered and undithered quantization is given, and the listening test and its results are described. The listening tests were made with 4- to 12-bits/sample and a sample rate of 44.1 kHz. The subjective tests show that subtractive dithering is preferred to undithered quantization up to 8-bits/sample. Nonsubtractive dithering was, with music stimuli, only preferred to undithered quantization at 4-bits/sample. With speech, nonsubtractive dither was preferred to undithered quantization up to 8-bits/sample. Triangular probability density function dither was only preferred to rectangular probability density function dither at very low bit rates.
Convention Paper 6328 (Purchase now)
Z1-2 DSP-Implemented Broadband Superdirective Microphone Array with Audible Noise Suppression—Jose-Luis Sánchez-Bote, Joaquin Gonzalez-Rodríguez, Javier Ortega-Garcia, Universidad Politecnica de Madrid - Madrid, Spain
In this paper a novel microphone linear array is proposed and implemented for real-time processing, working on a DSP processor in the frequency domain. The array, which is composed of 15 microphones in nested configuration, combines two multichannel techniques for speech improvement: SuperDirective beamforming (SD) and Audible Noise Suppression (ANS). The SD beamforming technique is an alternative to conventional or delay and sum beamforming (DS), which has worse low frequency spatial selectivity. ANS processing is based on the masking properties of the human auditory system and can benefit the perceived and objective quality of the processed signal. Although it has been successfully used in single channel systems, an enhanced multichannel version has been developed here, taking advantage of the extra information available in the acoustic spatial samples from the microphone array. Several on-line experiments are described here, assessing the real-time prototype.
Convention Paper 6329 (Purchase now)
Z1-3 Onset Detection, Music Transcription, and Ornament Detection for the Traditional Irish Fiddle—Aileen Kelleher, Dublin Institute of Technology - Dublin, Ireland; Derry Fitzgerald, Cork Institute of Technology - Cork, Ireland; Mikel Gainza, Eugene Coyle, Dublin Institute of Technology - Dublin, Ireland; Bob Lawlor, National University of Ireland - Maynooth, Ireland
By combining techniques used in previous onset detectors, a system that detects note onsets in traditional Irish fiddle tunes has been implemented. The notes detected also include the most common types of ornamentation played by the fiddle. Ornaments are notes of extremely short duration, at most a fifth the length of a regular note. A Sort Time Fourier Transform (STFT)-based subband technique, which previously gave good results for the Irish tin whistle, was modified to include a threshold approximation more suitable for the fiddle. This system has been tested on a database of real recorded fiddle tunes and good results have been achieved.
Convention Paper 6330 (Purchase now)
Z1-4 Perceptually Constrained Subspace Method for Enhancing Speech Degraded by Colored Noise—Adam Borowicz, Alexander Petrovsky, Bialystok Technical University - Bialystok, Poland
In this paper we present a novel method for enhancing speech corrupted by colored noise. A recent extension of a signal subspace approach to colored-noise processes is employed. Enhancement is performed using an optimal linear estimator, which minimizes average signal distortion power for a given set of constraints on the residual noise power spectrum. Perceptual criteria give lower speech distortion than SNR-based solutions. Thus, our proposition is to use constraints defined in a DFT domain that are consistent with masking properties of the human ear. The optimal filter is found by solving the constraints’ equations for the given masking threshold. The proposed method currently utilizes the most advanced ideas in signal subspace speech enhancement and is optimal in the general case of colored-noise process.
Convention Paper 6331 (Purchase now)
Z1-5 Blind Estimation of Room Resonances Using Popular, Classical, and Jazz Music—Takuya Yoshioka, NTT Corporation - Kyoto, Japan, and Kyoto University, Kyoto, Japan; Takafumi Hikichi, Masato Miyoshi, NTT Corporation - Kyoto, Japan; Hiroshi G. Okuno, Kyoto University - Kyoto, Japan
This paper describes a method for estimating the amplitude characteristics of room resonance modes from musical audio signals received by two microphones. Room resonance mode information is useful for many audio applications including soundfield equalization and control of the spatial aspects of timbre. Since the proposed method is characterized by blind processing, it does not need signals radiated from a source position. Simulation results obtained by using popular, classical, and jazz musical pieces showed that the proposed method could provide resonance frequencies and Qs, vital information for controlling the effects of resonances.
Convention Paper 6332 (Purchase now)
Z1-6 Comparison of Signal Reconstruction Methods for the Azimuth Discrimination and Resynthesis Algorithm—Dan Barry, Eugene Coyle, Dublin Institute of Technology - Dublin, Ireland; Bob Lawlor, National University of Ireland - Maynooth, Ireland
The Azimuth Discrimination and Resynthesis algorithm (ADRess), has been shown to produce high-quality sound source separation results for intensity panned stereo recordings. There are however, artifacts such as phasiness that become apparent in the separated signals under certain conditions. This is largely due to the fact that only the magnitude spectra for each of the separated sources are estimated. Each source is then resynthesized using the phase information obtained from the original mixture signal. This paper describes the nature and origin of the associated artifacts and proposes alternative techniques for resynthesizing the separated signals. A comparison of each technique is then presented.
Convention Paper 6333 (Purchase now)
Z1-7 Packet Loss Concealment for Audio Streaming Based on the GAPES Algorithm—Hadas Ofir, David Malah, Technion IIT - Haifa, Israel
In this paper we present a novel approach for audio packet loss concealment, designed for MPEG-Audio streaming, based only on the data available at the receiver. The proposed method is based on the GAPES (Gapped-data Amplitude and Phase Estimation) algorithm for replacing the missing data, using interpolation in the spectral domain. The MPEG standard uses the Modified Discrete Cosine Transform (MDCT) for compression. However, better interpolation results are obtained by converting the data to the Discrete Short-Time Fourier-Transform (DSTFT) domain. This conversion is done directly using an efficient procedure developed in this paper. This technique was tested subjectively and was found to provide better performance than previously reported papers, even with a packet loss rate of 30%.
Convention Paper 6334 (Purchase now)
Z1-8 An Accurate Method of Detection and Cancellation of Multiple Acoustic Feedbacks—Ariel F. Rocha, INESC Porto - Porto, Portugal; Aníbal Ferreira, University of Porto - INESC Porto, Porto, Portugal
This paper presents a new method to the adaptive cancellation of acoustic feedbacks. The method uses high resolution frequency analysis and high-Q notch filters so as to accurately detect feedbacks and cancel them without noticeably disturbing the main audio spectrum. The method will be described, its implementation on a TMS320C6711 DSP platform for real time operation will be explained, and results for the adaptive cancellation of two simultaneous acoustic feedbacks will be presented.
Convention Paper 6335 (Purchase now)
Z1-9 Fast Filterbanks for the Low Power MPEG High Efficiency Advanced Audio Coding Decoder—Shih-Way Huang, National Taiwan University - Taipei, Taiwan; Tsung-Han Tsai, National Central University - Chung-Li, Taiwan; Liang-Gee Chen, National Taiwan University - Taipei, Taiwan
The paper derives fast decomposition for the Quadrature Mirror Filter (QMF) banks of the low power Spectral Band Replication (SBR) tools in the MPEG High Efficiency Advanced Audio Coding (HE AAC) decoder. The original computation-intensive matrixing operations in the filterbanks are decomposed into fundamental Discrete Cosine Transform (DCT) types and simple permutations. Hence, the computational complexity can be effectively reduced by using fast algorithms for DCT.
Convention Paper 6336 (Purchase now)
©2005 Audio Engineering Society, Inc.