AES Berlin 2017
Paper Session P09
P09 - Spatial Audio—Ambisonics
Sunday, May 21, 09:00 — 12:30 (Salon 1 Moscow)
Franz Zotter (Chair)
P09-01 The Median-Plane Summing Localization in Ambisonics Reproduction
Bosun Xie (Presenting Author), Haiming Mai (Author), Xiaoli Zhong (Author)
One aim of Ambisonics reproduction is to recreate the perception of virtual source in arbitrary directions. Practical Ambisonics reproduction is unable to recreate correct high-frequency spectra in binaural pressures that are known as front-back and vertical localization cue. Based on both a simple shadowless head model and KEMAR’s HRTFs, the present work proves that the changes of ITD caused by head turning in Ambisonics match with these of a real source, and thus provide dynamic cue for vertical localization, especially in the median plane. In addition, the low-frequency virtual source direction can be approximately evaluated by using a set of localization equations or panning laws. The above analysis is validated by a median-plane virtual source localization experiment with Ambisonics reproduction.
Convention Paper 9726
P09-02 Exploring the Perceptual Sweet Area in Ambisonics
Matthias Frank (Presenting Author), Franz Zotter (Author)
A physical pressure-matching criterion to describe the size of the sweet area in Ambisonics does not quite match the generously large sweet area encountered in practice. To satisfy the need of a more practical characterization this paper comes up with a simple and systematic method to experimentally determine the perceptual sweet area. The method is not limited to assessing the localization of dry sounds as it also permits plausibility assessment of more complex scenes. This contribution exemplarily presents results for playback of two different audio scenes (dry/reverberant) using different Ambisonic orders.
Convention Paper 9727
P09-03 Phantom Source Widening by Filtered Sound Objects
Franz Zotter (Presenting Author), Matthias Frank (Author)
Audio effects increasing the perceived source extent (width/distance) often employ frequency-dependent panning of a single virtual sound object or real-time-controlled design of stochastic multichannel filters. Both ways imply increased complexity required in the renderer or object representation. In this paper we present a frequency-dependent panning scheme to obtain constellations of 3, 4, 5, or 7 filtered sound objects, as a simplified object-based description of wide/distant sound for any renderer. We deal with the multichannel filter-design questions: Are the filters rather temporally compact or frequency-selective, zero-phase FIR vs. IIR or causal-sided FIR, how strictly power-complementary? By results of a listening experiment for selected examples, we can provide some answers and an effective design of useful width-/distance-increasing filtered sound objects.
Convention Paper 9728
P09-04 Ambisonic Spatial Blur
Thibaut Carpentier (Presenting Author)
This paper presents a technique for controlling the spatial resolution of an Ambisonic sound field while preserving its overall energy. The proposed method allows to transform a stream encoded in N-order Ambisonic to a lower order resolution. The transformation can be continuously operated, indeed simulating fractional order representation of the Ambisonic stream and varying the "bluriness" of the spatial image.
Convention Paper 9729
P09-05 Comparing Ambisonic Microphones—Part 2
Enda Bates (Presenting Author), Francis M. Boland (Author), Sean Dooney (Author), Luke Ferguson (Author), Marcin Gorzel (Author), Hugh O'Dwyer (Author)
This paper presents some further experiments devised to assess the performance of an expanded number of commercially available Ambisonic microphones. The subjective timbral and spatial quality of five microphones (Soundfield MKV, Core Sound TetraMic, Sennheiser Ambeo, MH Acoustics Eigenmike, and Zoom H2N) is assessed using listening tests and a recording of an acoustic quartet. Localization accuracy is assessed using an objective directional analysis and recordings from a spherical loudspeaker array. Intensity vectors are extracted from 25 critical frequency bands of the Bark scale and used to compute the angle to the source location. Significant differences were found between microphones with the Soundfield MKV and Eigenmike producing the best results in terms of timbral quality and localization respectively.
Convention Paper 9730
P09-06 Object-Based Reverberation Encoding from First-Order Ambisonic RIRs
Philip Coleman (Presenting Author), Andreas Franck (Author), Philip J. B. Jackson (Author), Dylan Menzies (Author)
Recent work on a reverberant spatial audio object (RSAO) encoded spatial room impulse responses (RIRs) as object-based metadata that can be synthesized in an object-based renderer. Encoding reverberation into metadata presents new opportunities for end users to interact with and personalize reverberant content. The RSAO models an RIR as a set of early reflections together with a late reverberation filter. Previous work to encode the RSAO parameters was based on recordings made with a dense array of omnidirectional microphones. This paper describes RSAO parameterization from first-order Ambisonic (B-Format) RIRs, making the RSAO compatible with existing spatial reverb libraries. The object-based implementation achieves reverberation time, early decay time, clarity, and interaural cross-correlation similar to direct Ambisonic rendering of 13 test RIRs.
Convention Paper 9731
P09-07 Further Investigations on the Design of Radial Filters for the Driving Functions of Near-Field Compensated Higher-Order Ambisonics
Nara Hahn (Presenting Author), Sascha Spors (Author)
Analytic driving functions for Near-field Compensated Higher-order Ambisonics (NFC-HOA) are derived based on the spherical harmonics expansions of the desired sound field and the Green’s function that models the secondary sources. In the frequency domain, the radial part of the driving function is given as spherical Hankel functions and compensates the near-field effects of the secondary sources. By exploiting the polynomial expansion of the spherical Hankel functions, the radial filters can be implemented as cascaded biquad filters in the time domain, thereby reducing the computational complexity significantly. In this paper three practical issues regarding the design of the radial filters are addressed: pole-zero computation, pole-zero mapping, and gain normalization. Useful suggestions are given for improvements in terms of stability and numerical stability, which are demonstrated by numerical simulations.
Convention Paper 9732