AES San Francisco 2010
Paper Session P18
Saturday, November 6, 2:30 pm — 6:30 pm (Room 236)
Paper Session: P18 - Binaural Audio
Durand R. Begault
P18-1 Modification of HRTF Filters to Reduce Timbral Effects in Binaural Synthesis, Part 2: Individual HRTFs—Juha Merimaa, Sennheiser Research Laboratory - Palo Alto, CA, USA
In the first part of this study , a method for designing modified head-related transfer function (HRTF) filters with reduced timbral effects was proposed. Spectral localization cues were effectively scaled down while preserving the interaural time and level differences. For non-individualized HRTFs, the modifications were found to produce no statistically significant changes in localization. This paper continues the investigation using individual HRTFs. It is shown that in this case the reduction in timbral effects comes at a slight listener-dependent cost in localization performance. The filter design thus enables trading off more neutral timbre against more accurate localization.
Convention Paper 8265 (Purchase now)
P18-2 On the Improvement of Auditory Accuracy with Non-Indivisualized HRTF-Based Sounds—Catarina Mendonça, Jorge Santos, University of Minho - Minho, Portugal; Guilherme Campos, Paulo Dias, José Vieira, University of Aveiro - Aveiro, Portugal; João Ferreira, University of Minho - Minho, Portugal
Auralization is a powerful tool to increase the realism and sense of immersion in Virtual Reality environments. The Head Related Transfer Function (HRTF) filters commonly used for auralization are non-individualized, as obtaining individualized HRTFs poses very serious practical difficulties. It is therefore extremely important to understand to what extent this hinders sound perception. In this paper we address this issue from a learning perspective. In a set of experiments, we observed that mere exposure to virtual sounds processed with generic HRTF did not improve the subjects’ performance in sound source localization, but short training periods involving active learning and feedback led to significantly better results. We propose that using auralization with non-individualized HRTF should always be preceded by a learning period.
Convention Paper 8266 (Purchase now)
P18-3 Processing and Improving Head-Related Impulse Response Database for Auralization—Ben Supper, Focusrite Audio Engineering Ltd. - High Wycombe, UK
To convert a database of anechoic head-related impulse responses [HRIRs] into a set of data that is suitable for auralization involves many stages of processing. The output data set must be precisely corrected to account for some circumstances of the recording. It must then be equalized to remove coloration. Finally, the database must be interpolated to a finer resolution. This paper explains these stages of correction, equalization, and spatial interpolation for a frequently-used data set obtained from a KEMAR dummy head. The result is a useful database of HRIRs that can be applied dynamically to audio signals for research and entertainment purposes.
Convention Paper 8267 (Purchase now)
P18-4 Stimulus-Dependent HRTF Preference—Agnieszka Roginska, New York University - New York, NY, USA; Gregory H. Wakefield, University of Michigan - Ann Arbor, MI, USA; Thomas S. Santoro, Naval Submarine Medical Research Lab - Groton, CT, USA
Measurement of individual Head Related Transfer Functions (HRTFs) can be inconvenient, expensive, and time consuming. User selected HRTFs can alleviate the complexity of individually measured HRTFs and make better quality 3-D audio available to more listeners. This paper presents the results of a study designed to investigate the use of user-selected HRTFs augmented with customized interaural cues. In the study presented subjects were asked to select HRTFs that resulted in an accurate precept based on three specific criteria: externalization quality, elevation, and front/back discrimination. Subjective tests were conducted using three different stimuli. Results of the experiment are presented.
Convention Paper 8268 (Purchase now)
P18-5 Comparison between Spherical Headmodels and HRTFs in Upmixing for Headphone-Based Virtual Surround and Stereo Expansion—Part I—Sunil Bharitkar, Chris Kyriakakis, Audyssey Labs., Inc. - Los Angeles, CA, USA, University of Southern California, Los Angeles, CA, USA
In this paper, a first of multiple-parts, we compare the performance of headmodels with head-related transfer functions (HRTFs), which have previously been published, using different upmixing techniques for headphone virtual surround. We consider a spherical head, with and without the pinna or the torso model, whereas for the HRTFs we incorporate the CIPIC, Nagoya, and MIT HRTF sets in the up-mixing. The up-mixing technique includes using the Moorer reverberator, a modified Moorer reverberator, and modeling the direct sound, the first several discrete reflections (with adjustable delay and amplitude), and the diffuse field reflections with a tunable frequency dependent decorrelator. Furthermore, since the measured HRTFs can introduce audible coloration we investigate if there is a trade-off between localization and timbre by incorporating complex-domain smoothing of the HRTF time responses. To evaluate the localization and timbre performance between the models we use movie and music content (viz., stereo, ITU downmix, and a commercial down-mix method) as well as Gaussian tone noise bursts of critical bandwidth.
Convention Paper 8269 (Purchase now)
P18-6 HRTF Measurements with Recorded Reference Signal—Marko Durkovic, Florian Sagstetter, Klaus Diepold, Technische Universität München - München, Germany
Head-Related Transfer Functions (HRTFs) are used for adding spatial information in 3-D audio synthesis or for binaural robotic sound localization. Both tasks work best when using a custom HRTF database that fits the physiology of each person or robot. Usually, measuring HRTFs is a time consuming and complex procedure that is performed with expensive equipment in an anechoic chamber. In this paper we present a method that enables HRTF measurement in everyday environments by passively recording the surroundings without the need to actively emit special excitation signals. Experiments show that our method captures the HRTF's spatial cues and enables accurate sound localization.
Convention Paper 8270 (Purchase now)
P18-7 Angular Resolution Requirements for Binaural Room Scanning—Todd Welti, Harman International - Northridge, CA, USA; Xinting Zhang, State University of New York at Binghampton - Binghampton, NY, USA
Binaural Room Scanning is a method of capturing and reproducing a binaural representation of a room or car, using a dummy head incorporating binaural microphones, and individual measurements made with the dummy head positioned at a number of different head angles. The measurement process can be time-consuming. It is therefore important to know how high the angular resolution needs to be. An experiment was performed to see if the angular resolution could be reduced from the current 1 degree resolution to 15 degree resolution, without causing an audible difference. Using a 3 alternative forced choice method, trained listeners compared 1 degree and 15 degree angular resolution and could not reliably detect the difference.
Convention Paper 8271 (Purchase now)
P18-8 Binaural Reproduction of 22.2 Multichannel Sound over Loudspeakers—Kentaro Matsui, Akio Ando, NHK Science and Technology Research Laboratories - Tokyo, Japan
NHK has proposed the 22.2 multichannel sound system, which consists of 22 loudspeakers and 2 for LFE producing three-dimensional spatial sound, to be the format for future TV broadcasting. To allow it to be reproduced in homes, we have investigated various reproduction methods that use fewer loudspeakers. We introduce a design of binarual rendering of the 22.2 multichannel sound with three frontal loudspeakers as a minimum configuration model for homes. It can stably process the system inverse filters by dividing them into all-pass and minimum-phase components and successfully compensate the sound quality with a peak suppression method.
Convention Paper 8272 (Purchase now)