Listener's Head Movement Research Articles

Пространственная локализация цифрового звука в научном эксперименте и практике

<p style="text-align: justify;">Localization of sound in space is an important component of auditory perception, which is involved in the selection of various sound streams, the perception of speech in noise, and the organization of auditory images. Research over the past century has shown that sound localization is achieved through: differences in the intensity and time delay of sound waves arriving at different ears; spectral distortions arising from the anatomical features of the structure of the auricles, head, torso; dynamic cues (listener head movements), etc. However, some scientific and methodological issues (primarily related to the perception of natural sounds and the ecological validity of studies) have not been resolved. The development of digital audio techniques also leads to the emergence of new areas of research, including the processing of sound for the transmission of spatial information in headphones (which is solved using the head related transfer function — HRTF) and the creation of auditory interfaces. The tasks facing researchers in these areas are to improve the perception of spatial information (by manipulating the characteristics of the sound, prompts or training) and the creation of such sound events that can be perceived as object-related, i.e., inextricably linked with the purpose of the operator's activity. The methodology of the perceived quality of events, which makes it possible to distinguish which properties of the auditory image become the most important in human activity and which physical properties of the event they correspond to, can help in solving the tasks set and increasing the ecological validity of research.</p>

Updating of spatial selective auditory attention under-compensates for listener head movement

Listeners can use spatial selective auditory attention (SSAA) to focus on one talker in a complex acoustic scene. In our dynamic SSAA task, listeners oscillate their heads ~±40 deg at ~0.5 Hz while five different simultaneous sequences of four spoken digits are presented from loudspeakers at 0 deg azimuth (the target) and ±22.5 deg and ±45 deg azimuth (four distractors); listeners report the target sequence heard. We have observed [ASA, Minneapolis 2018] that listeners are more likely to misreport distractors that are centrally located in head-centered coordinates at the moment of presentation, and that under static conditions, performance declines with increasing target eccentricity—suggesting either that listeners cannot rapidly update the focus of their SSAA to compensate head motion (“lag” ) or that they have difficulty directing SSAA eccentrically (“low gain” ). To differentiate these alternatives, spatio-temporal maps of SSAA, conditioned on head orientation and direction of motion during each digit, were derived by computing the percentage of reported digits corresponding to those emitted by each loudspeaker. The spatial pattern of errors depended primarily on head orientation and not on direction of motion, suggesting that in this task SSAA tends to remain centrally focused in head-centered coordinates (low gain) rather than lagging dynamic head position.

Effects of vision, listener head movement, and target location on spatial selective auditory attention

Listeners can use spatial selective auditory attention (SSAA) to focus on one talker in a complex acoustic scene. We have reported [Macpherson & Ellis, ASA Salt Lake, 2016] that listeners’ ability to attend to a frontal target in the presence of spatially separated distractors is decreased in a head-movement condition, suggesting that listeners cannot rapidly update the focus of their SSAA to compensate fully for head motion. Participants reported anecdotally that being able to see the target loudspeaker seemed beneficial in directing SSAA. Here, we assessed the benefit of access to a visual reference frame in a similar task. Under lighted and dark conditions, on each trial, listeners either fixated toward 0 azimuth or oscillated their heads at ~0.5 Hz with an amplitude of ~±40 degrees while five different simultaneous equal-intensity sequences of four spoken digits were presented as target (at azimuths from 0 to 90 degrees) and four distractors (at azimuths ±22.5 and ±45 degrees relative to the central target). Listeners verbally reported the target sequence heard. Performance declined under head motion and with increasing target eccentricity, but not additionally in dark conditions, suggesting that a visual reference frame was not beneficial in directing SSAA in our task.Listeners can use spatial selective auditory attention (SSAA) to focus on one talker in a complex acoustic scene. We have reported [Macpherson & Ellis, ASA Salt Lake, 2016] that listeners’ ability to attend to a frontal target in the presence of spatially separated distractors is decreased in a head-movement condition, suggesting that listeners cannot rapidly update the focus of their SSAA to compensate fully for head motion. Participants reported anecdotally that being able to see the target loudspeaker seemed beneficial in directing SSAA. Here, we assessed the benefit of access to a visual reference frame in a similar task. Under lighted and dark conditions, on each trial, listeners either fixated toward 0 azimuth or oscillated their heads at ~0.5 Hz with an amplitude of ~±40 degrees while five different simultaneous equal-intensity sequences of four spoken digits were presented as target (at azimuths from 0 to 90 degrees) and four distractors (at azimuths ±22.5 and ±45 degrees relative to the central t...

A Stochastic Approximation Method with Enhanced Robustness for Crosstalk Cancellation

The objective of acoustic crosstalk cancellation is to use loudspeakers to deliver prescribed binaural signals (that reproduce a particular auditory scene) to a listener's ears, which is useful for 3-D audio applications. In practice, the actual transfer function matrix will differ from the design matrix, because of either the listener's head movement or rotation, etc. Crosstalk cancellation system (CCS) is very non-robust to these perturbations. Generally, in order to improve the robustness of CCS, several pairs of loudspeakers are needed whose position varies continuously as frequency varies. With the help of assumed stochastic analysis, we propose a stochastic robust approximation crosstalk cancellation method based on random perturbation matrix modeling the variations of the transfer function matrix. Under the free-field condition, simulation results demonstrate the effectiveness of the proposed method.

Effects of listener’s whole-body rotation and sound duration on sound localization accuracy

Listener's head movement is known to facilitate sound localization, which creates dynamic changes to the information input to both ears. For this study, we used a digitally controlled spinning chair to examine the effects of a listener’s whole-body rotation and sound duration on sound localization accuracy. We measured their sound localization accuracy at locations from left 30 deg to right 30 deg with respect to the listener. Stimuli were 1/3-octave band noise burst (fc = 1 kHz, SPL = 65 dB) with duration of 50, 200, and 1000 ms. Each stimulus was presented from a loudspeaker in a circular array. The listener, sitting on the chair at the circle center, reported the position of the presented stimulus in chair-still (0 deg/s) and chair-rotation (10 deg/s) conditions. Results showed superior sound localization accuracy of chair-rotation condition to that of a chair-still condition. Moreover, a significant effect of sound duration was observed, but interaction of the test condition and the sound duration was not significant. Although the listener's head movement might influence the localization performance more for long stimuli than for short stimuli, results suggest that the effects of a listener's whole-body rotation were less influenced by the sound duration.

The Effects of Hearing Impairment, Age, and Hearing Aids on the Use of Self-Motion for Determining Front/Back Location.

There are two cues that listeners use to disambiguate the front/back location of a sound source: high-frequency spectral cues associated with the head and pinnae, and self-motion-related binaural cues. The use of these cues can be compromised in listeners with hearing impairment and users of hearing aids. To determine how age, hearing impairment, and the use of hearing aids affect a listener's ability to determine front from back based on both self-motion and spectral cues. We used a previously published front/back illusion: signals whose physical source location is rotated around the head at twice the angular rate of the listener's head movements are perceptually located in the opposite hemifield from where they physically are. In normal-hearing listeners, the strength of this illusion decreases as a function of low-pass filter cutoff frequency, this is the result of a conflict between spectral cues and dynamic binaural cues for sound source location. The illusion was used as an assay of self-motion processing in listeners with hearing impairment and users of hearing aids. We recruited 40 hearing-impaired participants, with an average age of 62 yr. The data for three listeners were discarded because they did not move their heads enough during the experiment. Listeners sat at the center of a ring of 24 loudspeakers, turned their heads back and forth, and used a wireless keypad to report the front/back location of statically presented signals and of dynamically moving signals with illusory locations. Front/back accuracy for static signals, the strength of front/back illusions, and minimum audible movement angle were measured for each listener in each condition. All measurements were made in each listener both aided and unaided. Hearing-impaired listeners were less accurate at front/back discrimination for both static and illusory conditions. Neither static nor illusory conditions were affected by high-frequency content. Hearing aids had heterogeneous effects from listener to listener, but independent of other factors, on average, listeners wearing aids exhibited a spectrally dependent increase in "front" responses: the more high-frequency energy in the signal, the more likely they were to report it as coming from the front. Hearing impairment was associated with a decrease in the accuracy of self-motion processing for both static and moving signals. Hearing aids may not always reproduce dynamic self-motion-related cues with sufficient fidelity to allow reliable front/back discrimination.

Kinesic Patterning in Deceptive and Truthful Interactions

A persistent question in the deception literature has been the extent to which nonverbal behaviors can reliably distinguish between truth and deception. It has been argued that deception instigates cognitive load and arousal that are betrayed through visible nonverbal indicators. Yet, empirical evidence has often failed to find statistically significant or strong relationships. Given that interpersonal message production is characterized by a high degree of simultaneous and serial patterning among multiple behaviors, it may be that patterns of behaviors are more diagnostic of veracity. Or it may be that the theorized linkage between internal states of arousal, cognitive taxation, and efforts to control behavior and nonverbal behaviors are wrong. The current investigation addressed these possibilities by applying a software program called THEME to analyze the patterns of kinesic movements (adaptor gestures, illustrator gestures, and speaker and listener head movements) rated by trained coders for participants in a mock crime experiment. Our multifaceted analysis revealed that the quantity and quality of patterns distinguish truths from untruths. Quantitative and qualitative analyses conducted by case and condition revealed high variability in the types and complexities of patterns that were produced and differences between truthful and deceptive respondents questioned about a theft. Patterns incorporating adaptors and illustrator gestures were correlated in counterintuitive ways with arousal, cognitive load, and behavioral control, and qualitative analyses produced unique insights into truthful and untruthful communication.

The effect of listener head movement on perceived envelopment and apparent source width

Current research examining listeners' perceived spatial impression of a concert hall relies on a fixed-head worldview, since the overwhelming majority of listening tests conducted to determine subjective spaciousness [listener envelopment (LEV) and apparent source width (ASW)] has required listeners to keep their heads fixed. Such a worldview is an incomplete one, because listeners make noticeable exploratory head movements while evaluating sonic environments, including the more common task of source localization as well as the more involved task of evaluating the spaciousness of a concert hall. This study investigates the role of listener head movement in the evaluation of perceived LEV and ASW under 15 different concert hall conditions simulated over eight loudspeakers using Virtual Microphone Control. The conditions consist of both varying ratios of front-to-back energy and varying levels of cross-correlated reverberant energy. Head movements are monitored in terms of angular rotation (azimuth, elevation, and roll) using a head tracker while listeners are prompted to give subjective ratings of LEV and ASW ranging from 1 (least) to 7 (most). The listening tests are then repeated while subjects are asked to keep their heads fixed. The head movements are analyzed and results of the tests are compared.

Comparison of Japanese and English language descriptions of piano performances captured using popular multichannel microphone arrays

In a cross‐cultural comparison of musical sound evaluations, the way in which bipolar adjective pairs are used by native speakers of Japanese and English language was studied via a subjective rating task. These ratings were collected in response to eight solo piano performances that had been captured using four popular multichannel microphone arrays, reproduced via a standard 5‐channel loudspeaker array, re‐recorded binaurally, and finally presented via headphones. This allowed nearly identical stimuli to be presented to all listeners, without any modulation of the loudspeaker signals via listener head movements. Average ratings were compared to acoustical measures made on the 32 binaural stimuli, and to salient perceptual dimensions that previously had been derived from pairwise dissimilarity ratings between the stimuli. Results showed close agreement in how the selected terms were used by native speakers of Japanese and English language in the context of this study.

Acoustic rendering of a virtual environment based on virtual microphone control and binaural room scanning

Binaural room scanning (BRS) is a convolution technique that utilizes a set of spatially indexed impulse responses that are selected in response to listener head movements during virtual acoustic rendering, such that the direction of sonic elements is automatically updated according to the angle determined by a head tracker. Since the room impulse responses have to be measured for only a few loudspeaker positions, a very successful application of BRS is the simulation of a control room. In order to create a flexible headphone‐based virtual environment, it is proposed to simulate the input signals for the BRS system using virtual microphone control (ViMiC). ViMiC renders a virtual environment by simulating the output signals of virtual microphones that can be used to address a surround loudspeaker system. The advantages of this approach are twofold. First, the measured impulse responses of the BRS system ensure high spatial density in overall reflection patterns, avoiding the typical gaps in between the early reflections that occur when using solely mirror images or ray tracing. Second, the resulting imaging is closer to common audio engineering practice, where a pleasant but plausible environment is favored over a more purely realistic rendering.

PERFORMANCE EVALUATION OF MULTICHANNEL ADAPTIVE ALGORITHMS FOR LOCAL ACTIVE NOISE CONTROL

This paper deals with the development of a multichannel active noise control (ANC) system inside an enclosed space. The purpose is to design a real practical system which works well in local ANC applications. Moreover, the algorithm implemented in the adaptive controller should be robust, of low computational complexity and it should manage to generate a uniform useful-size zone of quite in order to allow the head motion of a person seated on a seat inside a car. Experiments were carried out under semi-anechoic and listening room conditions to verify the successful implementation of the multichannel system. The developed prototype consists of an array of up to four microphones used as error sensors mounted on the headrest of a seat place inside the enclosure. One loudspeaker was used as single primary source and two secondary sources were placed facing the seat. The aim of this multichannel system is to reduce the sound pressure levels in an area around the error sensors, following a local control strategy. When using this technique, the cancellation points are not only the error sensor positions but an area around them, which is measured by using a monitoring microphone. Different multichannel adaptive algorithms for ANC have been analyzed and their performance verified. Multiple error algorithms are used in order to cancel out different types of primary noise (engine noise and random noise) with several configurations (up to four channels system). As an alternative to the multiple error LMS algorithm (multichannel version of the filtered-X LMS algorithm, MELMS), the least maximum mean squares (LMMS) and the scanning error-LMS algorithm have been developed in this work in order to reduce computational complexity and achieve a more uniform residual field. The ANC algorithms were programmed on a digital signal processing board equipped with a TMS320C40 floating point DSP processor. Measurements concerning real-time experiments on local noise reduction in two environments and at frequencies below 230 Hz are presented. Better noise levels attenuation is obtained in the semianechoic chamber due to the simplicity of the acoustic field. The size of the zone of quiet makes the system useful at relatively low frequencies and it is large enough to cover a listener's head movements. The spatial extent of the zones of quiet is generally observed to increase as the error sensors are moved away from the secondary source, they are put closer together or its number increases. In summary, different algorithms' performance and the viability of the multichannel system for local active noise control in real listening conditions are evaluated and some guidelines for designing such systems are then proposed.

Masking in three-dimensional auditory displays.

The extent to which simultaneous inputs in a three-dimensional (3D) auditory display mask one another was studied in a simulated sonar task. The minimum signal-to-noise ratio (SNR) required to detect an amplitude-modulated 500-Hz tone in a background of broadband noise was measured using a loudspeaker array in a free field. Three aspects of the 3D array were varied: angular separation of the sources, degree of correlation of the background noises, and listener head movement. Masking was substantially reduced when the sources were uncorrelated. The SNR needed for detection decreased with source separation, and the rate of decrease was significantly greater with uncorrelated sources than with partially or fully correlated sources. Head movement had no effect on the SNR required for detection. Implications for the design and application of 3D auditory displays are discussed.