Optimal source distribution for binaural synthesis over loudspeakers

When binaural sound signals are presented with loudspeakers, the system inversion involved gives rise to a number of problems such as a loss of dynamic range and a lack of robustness to small errors and room reflections. The amplification required by the system inversion results in loss of dynamic range. The control performance of such a system deteriorates severely due to small errors resulting from, e.g., misalignment of the system and individual differences in the head related transfer functions at certain frequencies. The required large sound radiation results in severe reflection which also reduces the control performance. A method of overcoming these fundamental problems is proposed in this paper. A conceptual monopole transducer is introduced whose position varies continuously as frequency varies. This gives a minimum processing requirement of the binaural signals for the control to be achieved and all the above problems either disappear or are minimized. The inverse filters have flat amplitude response and the reproduced sound is not colored even outside the relatively large "sweet area." A number of practical solutions are suggested for the realization of such optimally distributed transducers. One of them is a discretization that enables the use of conventional transducer units.     

The time domain response of some systems for sound reproduction

The perception in a listener of the existence of a “virtual” source of sound at a prescribed spatial position can be produced by ensuring that the acoustic signals at the listener's ears faithfully replicate those that would be produced by a “real” source at the same position. When loudspeakers are used to transmit the signals, it is necessary to pass the signals intended for presentation at the listener's ears through a matrix of filters that provide the inverse of the matrix of transfer functions that relates the loudspeaker input signals to the listener's ear signals. The characteristics of such filter matrices are profoundly influenced by the conditioning of the matrix to be inverted. This filter design problem is reviewed here by representing the loudspeakers as simple point monopole sources the head of the listener as a rigid sphere. The case of a virtual acoustic imaging system that uses two loudspeakers in order to reproduce the signals at the two ears is first described in some detail and previous work is reviewed. It is confirmed that the time domain response of the reproduced field is of long duration at frequencies where the inversion problem is ill-conditioned. The influence of the presence of the listener's head on this time domain behaviour is also evaluated. The principle is then extended to four input-four output reproduction systems and the computational model is used to explain some previous experimental observations. Finally, the conditioning of five input-four output systems is also examined and shown to have some potentially desirable characteristics.     

前方空间环绕声的四扬声器虚拟重放

考虑头部转动带来的动态因素对听觉垂直定位的贡献,提出了前方空间环绕声的四扬声器虚拟重放方法。4个扬声器分别布置在水平面左前、右前以及高仰角的左前上、右前上方向,并采用听觉传输信号处理的方法将多通路空间环绕声信号转换为4个扬声器的重放信号。以9.1通路空间环绕声虚拟重放为例,采用头相关传输函数对双耳声压及其包含的定位因素进行分析表明,该方法可以产生正确的双耳时间差及其随头部转动的变化,从而产生合适的侧向定位双耳因素和垂直定位的动态因素。而心理声学实验结果表明,该方法可以重放稳定的前方空间的水平和垂直虚拟源。因此,四扬声器布置结合听觉传输处理足以重放前方空间环绕声的垂直定位信息,实现多通路空间环绕声的向下混合与简化。      

头部尺寸对虚拟声像定位的影响

本利用立体声像的相关理论分析方法，分析了用扬声器重发虚拟声时，倾听的头部尺寸对声像定位的影响。结果表明,当实际倾听的头部尺寸与虚拟声信号处理所用头部模型的尺寸有差别时，前方范围内的声像位置畸变较小，但侧向的声像位置畸变较大，因而本特别指出，倾听头部尺寸的不同是虚拟声重发时侧向的声像位置畸变的重要原因，声像定位实验证实了理论分析。     

Binaural active noise control using parametric array loudspeakers

This paper reports the binaural active noise control (ANC) system developed to deal with factory noise. The control points are located in the vicinity of the left and right ears of a worker sitting along the production line. Due to the complicated safety requirements in the factory, secondary sources and error microphones are not allowed to be placed near the worker. Therefore, the proposed ANC system employs the feedforward structure and adopts the parametric array loudspeakers (PALs) as the secondary sources. The PAL is a type of directional loudspeaker that generates a much narrower sound field as compared to the conventional loudspeaker. Once the proposed ANC system has been trained offline, the error microphones can be removed. The performance of the binaural ANC system is successfully demonstrated based on a digital signal processor (DSP) implementation.     

Spatial sound resolution of an interpolated HRIR library

The paper evaluates the human directional resolution of virtual sound sources synthesised with the aid of a generalised head related impulse response (HRIR) library, i.e., an HRIR library measured using a dummy head and torso. The original HRIR set is first expanded using linear interpolation, and then directional resolution measurements are performed for playback through headphones. These results are compared to the results obtained using loudspeakers as sound sources in an anechoic chamber. Directional resolution is the ability of listeners to distinguish two closely-spaced sound sources alternately playing the same signal. Experiments show that two sound sources with insufficient spacing appear as a single source to the listener. Directional resolution for small azimuth changes is relatively high for both virtual and real sound sources. Most test subjects have no problem resolving two sound sources only 5° apart. Compared to real sound sources, detecting changes in elevation of virtual sound sources is much less accurate, which may be the main drawback of using a generalised HRIR library.     

两扬声器虚拟声重放的双耳声压控制与定位性能

在两扬声器虚拟声重放中,通过精确重构双耳声压而产生不同的空间听觉感知。其重放的定位性能应该是由双耳声压控制的代价和稳定性所共同决定的。过去研究主要对双耳声压控制的稳定性进行分析,并以此作为扬声器布置和信号处理的依据。该文研究表明仅对双耳声压的稳定性分析是不足以完全衡量扬声器虚拟声重放的定位性能的。进一步采用虚拟声信号处理滤波器响应平均功率对双耳声压控制的代价进行分析。结果表明,缩窄左右对称扬声器布置的张角或采用非对称扬声器布置会明显增加产生侧向目标虚拟源时的双耳声压控制代价。虚拟源(虚拟声像)定位实验表明,双耳声压控制代价增加会引起虚拟源定位缺陷。实际应用中,为了有效产生侧向虚拟源,应避免采用过窄张角(如立体声偶极)和非对称的扬声器布置。     

A Comparison on the Localization Performance of Static and Dynamic Binaural Ambisonics Reproduction with Different Order

Ambisonics is a series of spatial sound reproduction system based on spatial harmonics decomposition and each order approximation of sound field. Ambisonics signals are originally intended for loudspeakers reproduction. By using head-related transfer functions (HRTFs) filters, binaural Ambisonics converts the Ambisonics signals for static or dynamic headphone reproduction. In present work, the performances of static and dynamic binaural Ambisonics reproduction are evaluated and compared. The mean binaural pressure errors across target source directions are first analyzed. Then a virtual source localization experiment is conducted, and the localization performances are evaluated by analyzing the percentages of front-back and up-down confusion, the mean angle error and discreteness in the localization results. The results indicate that binaural Ambsonics reproduction with insufficiently high order (for example, 5-10 order) is unable to recreate correct high-frequency magnitude spectra in binaural pressures, resulting in degradation in localization for static reproduction. Because dynamic localization cue is included, dynamic binaural Ambisoncis reproduction yields obviously better localization performance than static reproduction with the same order. Even a 3-order dynamic binaural Ambisoncis reproduction exhibits appropriate localizations performance.     

Sound localization of stereo reproduction with parametric loudspeakers

A parametric loudspeaker radiates an audible signal by the interaction of the primary wave that is amplitude modulated and is known as a super-directivity loudspeaker. The parametric loudspeaker is one of the prominent applications of nonlinear acoustics. So far, the applications have been limited monaural reproduction sound system for public address in museum, station, street etc. In this paper, we investigated sound localization of stereo reproduction using two parametric loudspeakers in comparison with that using two ordinary dynamic loudspeakers. In subjective tests, the binaural information ILD (Interaural Level Difference) or ITD (Interaural Time Delay) was focused on. To investigate the characteristics of sound localization in a wide listening area, three typical listening positions were picked up. Signals were 500 Hz, 1 kHz, 2 kHz and 4 kHz pure tones and pink noise. The used parametric loudspeaker was an equilateral hexagon. The subjective test led to the results that when the parametric loudspeakers were used, the listeners at the three typical listening positions perceived the correct sound localization of not only pure tone but also pink noise and when the ordinary dynamic loudspeakers were used, except for the case of pure tone with ITD, the tendency was almost similar to those using the parametric loudspeakers. The second subjective tests were conducted in order to investigate in details the difference between parametric loudspeakers and ordinary dynamic loudspeakers by increasing the number of subjects. In the case of ITD and 500 Hz using the ordinary dynamic loudspeakers, three types of sound localization were categorized, in which the reversed type was major and the normal and the other types were minor. The ILDs which were measured with a dummy head and were calculated with several formulas were almost the same and indicated the reasons of the reversed typed sound localization and a serious influence of the crosstalk. It was found that in the case of pure tone with ITD, the contradiction between the binaural information ILD and ITD is remarkable, because the directivity of the ordinary dynamic loudspeakers was so dull that the crosstalk components had a serious influence on sound localization. It was determined the parametric loudspeaker could transmit correct binaural information to the listener, because the directivity of the parametric loudspeakers was so sharp that it suppressed the cross talk components.     

The effects of spatial separation in distance on the informational and energetic masking of a nearby speech signal

Although many studies have shown that intelligibility improves when a speech signal and an interfering sound source are spatially separated in azimuth, little is known about the effect that spatial separation in distance has on the perception of competing sound sources near the head. In this experiment, head-related transfer functions (HRTFs) were used to process stimuli in order to simulate a target talker and a masking sound located at different distances along the listener's interaural axis. One of the signals was always presented at a distance of 1 m, and the other signal was presented 1 m, 25 cm, or 12 cm from the center of the listener's head. The results show that distance separation has very different effects on speech segregation for different types of maskers. When speech-shaped noise was used as the masker, most of the intelligibility advantages of spatial separation could be accounted for by spectral differences in the target and masking signals at the ear with the higher signal-to-noise ratio (SNR). When a same-sex talker was used as the masker, the intelligibility advantages of spatial separation in distance were dominated by binaural effects that produced the same performance improvements as a 4-5-dB increase in the SNR of a diotic stimulus. These results suggest that distance-dependent changes in the interaural difference cues of nearby sources play a much larger role in the reduction of the informational masking produced by an interfering speech signal than in the reduction of the energetic masking produced by an interfering noise source.     

Binaural unmasking in infants

Localization responses to a broadband noise signal presented against a broadband noise masker were obtained from 12-month-old infants and adults. Two loudspeakers, one to the left and one to the right of the listener, continuously presented identical broadband maskers. On a trial, a broadband signal was added to one of the loudspeakers. Subjects were required to identify the loudspeaker producing the signal. Noise signals were either coherent (from the same noise generator) or incoherent (from an independent noise generator). Both infants and adults found it easier to locate the incoherent signals even when the two types of signals were adjusted to produce equal increments in power. Since monaural performance, after this adjustment, should be equivalent for the two cases, superior performance for incoherent signals implies that binaural processing is involved. The same result was observed in control experiments in which coherent and incoherent signals were presented over earphones to adults. These results suggest that the mechanisms responsible for binaural unmasking are operative by 12 months of age.     

双耳效应机理教学实验设计

应用Cool Edit Pro2.0声音处理软件生成声源样本文件,采用仿真人工头进行不同方位的声音录制,并应用Cool Edit Pro2.0对声音录制信号进行后期处理,设计了双耳效应中相位差、声级差的定量实验,时间差的半定量和音色差的定性实验.     

Separation of nonstationary sound fields in the time-wavenumber domain

A number of sound field separation techniques have been proposed for different purposes. However, these techniques just consider the separation of sound fields in the space domain and are restricted to stationary sound fields. When the sound fields are nonstationary, it is also necessary to perform the separation in the time domain. Therefore, on the basis of the propagation principle of sound pressure in the time-wavenumber domain, a nonstationary sound field separation technique with two closely spaced parallel measurement surfaces is proposed. It can separate the nonstationary signals generated by the primary sources in both time and space domains when the disturbing sources exist on the other side of the measurement plane. The signals in time and space domains are separated by using the spatial Fourier transform method and the time domain deconvolution method. A simulation involving two monopoles driven by nonstationary signals demonstrates that the method proposed can remove the influence of disturbing sources in both time and space domains. The feasibility of this method is also demonstrated by an experiment with two loudspeakers located on two sides of measurement planes. Additionally, to comment more objectively on the separation results, some indicators are computed in both the simulation and experiment.     

重放自由场虚拟源距离信息的动态双耳Ambisonics方法

双耳重放的目标之一是在耳机重放中产生不同方向和距离的虚拟源感知。本文研究了动态双耳Ambisonics重放自由场虚拟源方向和距离信息的简化信号处理方法。该信号处理方法包括两步:第1步是基于目标声场的球谐函数分解,合成采用扬声器的近场Ambisonics重放中逐级重构目标声场的信号;第2步是采用虚拟扬声器重放的方法,用动态头相关函数滤波处理将Ambisonics的扬声器重放信号转换为双耳重放信号并用耳机重放。进一步研究了动态双耳Ambisonics的阶数对定位效果的影响,为简化信号处理提供依据。对重放产生的双耳声压分析表明,5阶动态双耳Ambisonics重放足以提供听觉方向定位和距离感知的重要信息。同时心理声学的实验结果表明,结合声源距离相关的响度因素,5阶动态双耳Ambisonics重放可产生不同方向和1.0 m以下不同近场距离的自由场虚拟源的听觉感知。本文的方法仅需要固定距离的48个均匀空间方向的远场非个性化HRTF处理,实现了信号处理的简化。      

The contribution of the near and far ear toward localization of sound in the sagittal plane

Eight listeners were required to locate a train of 4.5-kHz high-pass noise bursts emanating from loudspeakers positioned +/- 30, +/- 20, +/- 10, and 0 deg re: interaural axis. The vertical array of loudspeakers was placed at 45, 90, and 135 deg left of midline. The various experimental conditions incorporated binaural and monaural listening with the latter utilizing the ear nearest or ear farthest from the sound source. While performance excelled when listening with only the near ear, the contribution of the far ear was statistically significant when compared to localization performance when both ears were occluded. Based on head related transfer functions for stimuli whose bandwidth was 1.0 kHz, four spectral cues were selected as candidates for influencing location judgments. Two of them associated relative changes in energy across center frequencies (CFs) with vertical source positions. The other two associated an absolute minimum (maximum) energy for specific CFs with a vertical source position. All but one cue when measured for the near ear could account for localization proficiency. On the other hand, when listening with the far ear, maximum energy at a specific CF outperformed the remaining cues in accounting for localization proficiency.     

Optimal design of loudspeaker arrays for robust cross-talk cancellation using the Taguchi method and the genetic algorithm

An optimal design technique of loudspeaker arrays for cross-talk cancellation with application in three-dimensional audio is presented. An array focusing scheme is presented on the basis of the inverse propagation that relates the transducers to a set of chosen control points. Tikhonov regularization is employed in designing the inverse cancellation filters. An extensive analysis is conducted to explore the cancellation performance and robustness issues. To best compromise the performance and robustness of the cross-talk cancellation system, optimal configurations are obtained with the aid of the Taguchi method and the genetic algorithm (GA). The proposed systems are further justified by physical as well as subjective experiments. The results reveal that large number of loudspeakers, closely spaced configuration, and optimal control point design all contribute to the robustness of cross-talk cancellation systems (CCS) against head misalignment.     

Investigation of spherical loudspeaker arrays for local active control of sound

Active control of sound can be employed globally to reduce noise levels in an entire enclosure, or locally around a listener's head. Recently, spherical loudspeaker arrays have been studied as multiple-channel sources for local active control of sound, presenting the fundamental theory and several active control configurations. In this paper, important aspects of using a spherical loudspeaker array for local active control of sound are further investigated. First, the feasibility of creating sphere-shaped quiet zones away from the source is studied both theoretically and numerically, showing that these quiet zones are associated with sound amplification and poor system robustness. To mitigate the latter, the design of shell-shaped quiet zones around the source is investigated. A combination of two spherical sources is then studied with the aim of enlarging the quiet zone. The two sources are employed to generate quiet zones that surround a rigid sphere, investigating the application of active control around a listener's head. A significant improvement in performance is demonstrated in this case over a conventional headrest-type system that uses two monopole secondary sources. Finally, several simulations are presented to support the theoretical work and to demonstrate the performance and limitations of the system.     

扬声器的特性不匹配对虚拟声像的影响

本文从理论和实验上探讨了扬声器的特性不匹配对重放虚拟声像的影响。结果表明,用两扬声器进行虚拟听觉重放时,在某些频率段和虚拟声像角度,两扬声器间很小的幅频特性差异或相频特性差异都足以对虚拟声像方向产生明显的影响。扬声器特性的差异对前方范围的声像影响较小,但对侧向范围的声像影响较大,因而两扬声器的特性不匹配也是导致虚拟听觉重放时侧向声像位置畸变的重要原因之一。而在实际应用中要特别注意两扬声器的特性匹配,或者要用信号处理的方法对两扬声器的特性进行校正。     

Evaluation of array-processing algorithms for a headband hearing aid

Several array-processing algorithms were implemented and evaluated with experienced hearing-aid users. The array consisted of four directional microphones mounted broadside on a headband worn on the top of the listener's head. The algorithms included two adaptive array-processing algorithms, one fixed array-processing algorithm, and a reference condition consisting of binaural directional microphones. The algorithms were evaluated under conditions with both one and three independent noise sources. Performance metrics included quantitative speech reception thresholds and qualitative subject preference ratings for ease-of-listening measured using a paired-comparison procedure. On average, the fixed algorithm improved speech reception thresholds by 2 dB, while the adaptive algorithms provided 7-9-dB improvement over the reference condition. Subjects judging ease-of-listening generally preferred all array-processing algorithms over the reference condition. The results suggest that these adaptive algorithms should be evaluated further in more realistic acoustic environments.     

The impact of early reflections on binaural cues

Animals live in cluttered auditory environments, where sounds arrive at the two ears through several paths. Reflections make sound localization difficult, and it is thought that the auditory system deals with this issue by isolating the first wavefront and suppressing later signals. However, in many situations, reflections arrive too early to be suppressed, for example, reflections from the ground in small animals. This paper examines the implications of these early reflections on binaural cues to sound localization, using realistic models of reflecting surfaces and a spherical model of diffraction by the head. The fusion of direct and reflected signals at each ear results in interference patterns in binaural cues as a function of frequency. These cues are maximally modified at frequencies related to the delay between direct and reflected signals, and therefore to the spatial location of the sound source. Thus, natural binaural cues differ from anechoic cues. In particular, the range of interaural time differences is substantially larger than in anechoic environments. Reflections may potentially contribute binaural cues to distance and polar angle when the properties of the reflecting surface are known and stable, for example, for reflections on the ground.