• Sessions by Industry
• Detailed Calendar
• Convention Planner
• Paper Sessions
• Master Classes
• Live Sound Seminars
• Exhibitor Seminars
• Special Events
• Student Program
• Technical Tours
• Technical Council
• Standards Committee
• Heyser Lecture
AES Amsterdam 2008
Signal Processing, Sound Quality Design
Poster Session P13
Sunday, May 18, 16:00 — 17:30
P13-1 Time-Alignment of Multiway Loudspeakers with Group Delay Equalization: Part I—Sunil Bharitkar, Chris Kyriakakis, Tom Holman, University of Southern California and Audyssey Labs - Los Angeles, CA, USA
In this paper, a first of two-parts, a technique for time-aligning the driver responses (viz., woofer, mid-range, and tweeter responses) in a multiway loudspeaker system is presented. Generally, woofers exhibit a much larger time-of-arrival delay at a listening position, compared to the mid-range and high-frequency drivers due to the presence of crossover networks. Moreover, the time-of-arrival delay for all drivers is frequency dependent exhibiting a large variation over the audible frequency domain. Due to these differences a two-part study was undertaken to understand the effects of these variations, quantitatively and qualitatively, in the direct as well as reverberant field in typical listening rooms with diverse content. Various multi-way loudspeakers, measured in one of the anechoic chambers at Audyssey, were selected to provide a diverse corpus of responses. In this first part we present the motivation behind the system used for applying all-pass filters to process audio signals being delivered to the multiway speaker and propose a time-delay difference equalization technique, between the drivers of the multiway loudspeakers showing group-delay equalization while retaining a flat magnitude response. Clearly, applying all-pass filters will result in temporal-smearing of the measured response. Furthermore, an investigation using perceptually motivated variable-octave complex smoothing of responses, and designing all-pass filters based on this phase-smoothed data, will also be undertaken. Quantitative results obtained will be presented in this paper whereas the next part of the two-part paper will present results from listening tests.
Convention Paper 7396 (Purchase now)
P13-2 Singing Voice Separation Combining Panning Information and Pitch Tracking—Maximo Cobos, Jose J. Lopez, Technical University of Valencia - Valencia, Spain
Source Separation techniques applied to music mixtures are able to extract relevant information that can be very useful for many applications, such as music remixing and reprocessing, lyrics recognition or music information retrieval. Among all the sources present in modern music themes, the singing voice has a especial interest because it is the only one that combines music, lyrics, and expression. In this paper we propose a system designed for extracting the singing voice from stereo recordings in different steps. This system combines panning information and pitch tracking, allowing the refinement of the time-frequency mask applied for extracting a vocal segment, and thus, improving the separation. An application example is discussed.
Convention Paper 7397 (Purchase now)
P13-3 The Downsampling Dilemma: Perceptual Issues in Sample Rate Reduction—Brett Leonard, New York University - New York, NY, USA
Many options currently exist for sample rate conversion. With sample rate reduction playing an integral part in the modern production world, downsampling algorithm quality is more important than ever. This paper presents data exploring the differences in sample rate reduction algorithms. While certain tests clearly display differences in the quality of the algorithms, listening test data shows the average listener is unable to repeatedly discern the difference in sample rate reduction methods.
Convention Paper 7398 (Purchase now)
P13-4 NU-Tech: The Entry Tool of the hArtes Toolchain for Algorithms Design—Ariano Lattanzi, Ferruccio Bettarelli, Leaff Engineering - Porto Potenza Picena (MC), Italy; Stefania Cecchi, Univerista Politecnica delle Marche - Ancona, Italy
The aim of the hArtes project is to facilitate and automate the rapid design and development of heterogeneous embedded systems, targeting a combination of a general purpose embedded processor, digital signal processing, and reconfigurable hardware. In this paper we present the NU-Tech platform, the main entry tool from the hArtes toolchain, which has the role of assisting the designers in tuning and possibly improving the input algorithm at the highest level of abstraction. A brief description of the project itself will be given and its vocation to audio highlighted through a case study application.
Convention Paper 7399 (Purchase now)
P13-5 Recovery of Missing Signals Utilizing (GHA) Generalized Harmonic Analysis—Applied Interpolation—Teruo Muraoka, Takahiro Miura, Tohru Ifukube, University of Tokyo - Tokyo, Japan
For archiving damaged historical recordings, recoveries of missing portions are as essentially important as noise reduction. Conventional counter-measures with functional interpolation are not effective when the missing interval is long. Inharmonic frequency analysis GHA is profitable for this purpose, because the recomposed signal with frequency components obtained by GHA exhibits very long periods. Length of the period is given as an inverse number of the least common multiple of the rendered inharmonic frequency’s periods. This feature is very advantageous for signal recovery, and the authors devised an extrapolation simply extending a re-synthesizing waveform obtained through GHA analysis/re-synthesis. The authors got satisfactory results as a whole by applying interpolation combined with forward and backward extrapolations based upon the abovementioned method. Results of recovery highly depend upon characters of signals (such as music), and the authors did not find definite rules for setting GHA’s analyzing conditions. Those are given through auditory examination this time.
Convention Paper 7400 (Purchase now)
P13-6 Combination of Warped and Linear Filter Structures for Loudspeaker Equalization—German Ramos, Jose J. Lopez, Technical University of Valencia - Valencia, Spain; Basilio Pueo, University of Alicante - Alicante, Spain
The warping filters were introduced years ago for loudspeaker equalization in order to solve the lack of resolution of the linear filters at low frequencies, and also to follow the frequency resolution of psycho-acoustic scales like the Bark scale, with a more logarithmic than linear behavior. However, this improvement in the frequency resolution at low and mid frequencies is done at the expense of loosing resolution at high frequencies and increasing the complexity of the filter and its implementation computational cost. In this paper a smart combination of linear and warped filter structures previously developed by the authors for FIR filters is presented with new contributions and extended to IIR filters. This combination saves computational cost and obtains a proper frequency resolution at the whole frequency band, obtaining better results for the same computational cost than when using linear or warped filters alone. The results have been subjectively tested using the ABX methodology with successfully results. The presented filter structures, methodology, and apparatus to do the filtering are patent pending.
Convention Paper 7401 (Purchase now)
P13-7 Multichannel Dereverberation System Using Modified Correlation-Based Blind Deconvolution and Multi-Microphone Spectral Subtraction—Jae-woong Jeong, Yonsei University - Seoul, Korea; Young-cheol Park, Yonsei University - Wonju, Korea; Seok-Pil Lee, Korea Electronics Technology Institute (KETI) - Sungnam, Korea; Dae-hee Youn, Yonsei University - Seoul, Korea
This paper presents a new multichannel dereverberation system combining modified correlation-based blind deconvolution with multi-microphone spectral subtraction. In the proposed system, we make M combinations of observed signals and apply them to the correlation-based blind deconvolution. The deconvolved signals are then used as inputs to the multi-microphone spectral subtraction. These spectral subtractions with the multiple deconvolved signals estimate the reverberant energy by using both a frame delay and a frequency-dependent weight. Due to the accurate estimation of the reverberant energy, the combination of correlation-based blind deconvolution with the multi-microphone spectral subtraction provides improved dereverberation performance. Performance improvement of the proposed system has been confirmed through experiments.
Convention Paper 7402 (Purchase now)
P13-8 Harmonic and Intermodulation Analysis of Nonlinear Devices Used in Virtual Bass Systems—Nay Oo, Woon-Seng Gan, Nanyang Technological University - Singapore
Nonlinear Devices (NLD) are used in virtual bass system. NLD generates harmonics which in turn create the pitch perception and are used in audio bass enhancement systems using psychoacoustics. This paper presents the mathematical derivations and analysis of five different NLD devices, together with intermodulation analysis of harmonics generated by these NLDs. The five NLDS are half-wave rectifier, full-wave rectifier, square wave, polynomial function, and exponential function. The derivation of harmonic analysis equations are based on Fourier Theorems, Chebyshev Polynomials, and Taylor Series expansions. Besides the harmonics, intermodulation components also resulted from NLDs. Both mathematical analysis and simulation results are presented for the intermodulation effects of harmonics generated by NLDS.
Convention Paper 7403 (Purchase now)
Last Updated: 20080612, tendeloo