Authors:Romblom, David; Depalle, Philippe; Guastavino, Catherine; King, Richard
Affiliation:McGill University, Montréal, Québec, Canada; Centre for Interdisciplinary Research in Music Media and Technology (CIRMMT), Montréal, Québec, Canada
A reverberant diffuse sound field is characterized by incoherent energy arriving from all directions and is perceptually described as an auditory event that is heard everywhere. It is common practice for sound recording engineers to use differing microphone strategies for the direct and diffuse fields. While there are a variety of techniques to record and reproduce point sources, a systematic tool for diffuse sound fields does not exist. Diffuse Field Modeling (DFM) is a physically-inspired method for approximating a diffuse field in order to create a natural-sounding room effect for arbitrary loudspeaker configurations. It is intended to function in parallel with point source techniques. Using a statistical description of reverberation, the decorrelation filters in DFM are based on physical acoustics, and the resulting diffuse fields are validated with simulations incorporating the Kirchhoff/Helmholtz Integral. The resulting diffuse fields have the expected spatial autocorrelation, and the channels of the array have the expected frequency-dependent correlation. The filters can be tuned to introduce random variation that has physically-plausible frequency autocorrelation, which strongly influences the spatial impression.
Download: PDF (HIGH Res) (40.1MB)
Download: PDF (LOW Res) (897KB)
Authors:Rummukainen, Olli; Romblom, David; Guastavino, Catherine
Affiliation:Aalto University, School of Electrical Engineering, Espoo, Finland; McGill University, Montreal, Quebec, Canada; Centre for Interdisciplinary Research in Music Media and Technology (CIRMMT), Montreal, Quebec, Canada
The Diffuse Field Model (DFM) described in Part 1 is perceptually evaluated in this article. Two experiments were conducted. In first experiment, sound recording professionals rated different treatments of DFM presented on a 20-channel array. This evaluation included the geometric modeling of reflections, strategies involving the early portion of the B-Format Room Impulse Response (RIR), and a comparison between 0th- and 1st-order RIR. Results indicate that it is necessary to model the earliest reflections and to use all four channels of the B-Format room impulse response. In the second experiment, musicians and sound recording professionals were asked to rate DFM and common microphone techniques presented on 3/2 stereophonic setup. DFM was found to be perceptually comparable to the Hamasaki Square technique. DFM approach used in this study is part of a physically-plausible virtual acoustic model for sources that were captured with close microphone placement. This model replaces the panning, delay, and reverberation that would typically be used. DFM is a perceptually viable method to create room impression that allows free placement of anechoic point sources in arbitrary multichannel loudspeaker setups.
Download: PDF (HIGH Res) (19.2MB)
Download: PDF (LOW Res) (429KB)
Authors:Storek, Dominik; Bouse, Jaroslav; Rund, Frantisek; Marsalek, Petr
Affiliation:Department of Radioelectronics, Czech Technical University in Prague, Prague, Czech Republic
This article introduces three methods for artifact reduction in Differential Head-Related Transfer Function (DHRTF). Such artifacts may occur within virtual sound source positioning methods when spike-like spectral peaks occur in the module of the DHRTF. These spectral features result in an undesirable whistling-like sound component that distorts both timbre and spatial perception of the virtual sound. DHRTF filters require preprocessing by an appropriate method that is able to remove or reduce the artifacts while preserving both the natural character of the timbre and the spatial perception of the positioned sound. Preprocessing of the DHRTF set avoids the effects of the negative ILD while preserving all the necessary localization cues and natural timbre of the positioned sound. Three methods based on limiting and/or smoothing of the DHRTF magnitude by a low-pass filter were introduced and evaluated by both paired comparison listening tests and objective assessment techniques. Even elementary preprocessing algorithms improve the quality of the positioned sound.
Download: PDF (HIGH Res) (4.3MB)
Download: PDF (LOW Res) (440KB)
Affiliation:Purdue School of Engineering-IUPUI, Indianapolis, Indiana
To understand how audio recording and production (ARP) programs meet the needs of the audio industry, the New Hires Survey (NHS) study asked new hires to articulate the skills they have acquired, to rate their proficiency and to indicate where they learned them. New hires were between the ages of 21 and 30, and attended 3- and 4-year professional schools or 4-year music colleges for their formal ARP training. They were employed as live sound/recording engineers at medium to large businesses located in different geographic regions of the U.S. While they learned basic technical skills during formal ARP training, they learned social and communication skills by themselves or on the job. New hires requested a greater emphasis on career critical areas of the live sound and music business. Further curriculum design and research is recommended to understand industry needs, identify best practices for the acquisition of skills, and to determine how educational institutions can keep pace with the ever-changing audio industry. Meeting the needs of the audio industry is imperative for the success of these programs and their graduates.
Download: PDF (HIGH Res) (1.0MB)
Download: PDF (LOW Res) (175KB)
Authors:Anderson, David; Bocko, Mark F.
Affiliation:University of Rochester, Rochester, NY, USA
Loudspeakers that produce sound from the bending vibrations of flat plates offer an attractive design alternative for applications in which a low-profile loudspeaker form factor is desirable, such as displays or windows. However, vibrating panel speakers may exhibit irregular low-frequency response and phase distortions due to the presence of isolated low-frequency panel modes. A method is presented for tuning the frequency response of flat-panel loudspeakers by employing force-driver arrays. The method enables independent actuation of specific panel bending modes, which can be combined with a frequency crossover network to allow extensive tuning of the loudspeaker. Experiments demonstrate that independent control of plate modes is achievable with a small array of inertial drivers affixed to the plate and a wide variety of acoustic responses are possible, including cancellation of the adverse effects of isolated low-frequency plate modes.
Download: PDF (HIGH Res) (5.4MB)
Download: PDF (LOW Res) (564KB)
Authors:Iversen, Niels Elkjær; Knott, Arnold; Andersen, Michael A. E.
Affiliation:Technical University of Denmark, Kgs. Lyngby, Denmark
For decades voice coils have been designed for a nominal resistance of 3 to 8 ohms, despite the fact that modern audio amplifiers using switch-mode technology can be easily handle much lower impedance loads. A thorough analysis of loudspeaker efficiency showed that the efficiency can be expressed as a function of the voice coil fill factor and the geometry of the magnet system. In addition, high mass ratios are more beneficial for the efficiency of drivers using high fill factor voice coils than drivers with low mass ratio. Different voice coil winding strategies are described and their fill factors analyzed. It is found that by lowering the nominal resistance of a voice coil using rectangular wire, one can increase the fill factor. A higher fill factor will shift the low frequency upward, resulting in higher -3 dB cut-off frequencies. By using rectangular wire with low DC resistance, the fill factor could be significantly increased. The fill factor of a conventional 4-ohm voice coil was measured to be 53%.
Download: PDF (HIGH Res) (7.5MB)
Download: PDF (LOW Res) (368KB)
Fifty years of artificial reverberation have brought about surprisingly convincing algorithms capable of simulating natural acoustics effectively, but machines still find it difficult to determine “how much is enough” when attempting to perform automatic audio mixing. Removing reverberation from a mixture without knowing much about the original signal is a hard challenge, but various people have made inroads using methods brought from blind source separation and echo suppression. Humans are still considerably better at stream segregation than machines. Machines, however, may be able to employ many more “ears” to achieve spatial selectivity.
Download: PDF (251KB)