Activity

Regional variation in American English speech is often described in terms of shifts, indicating which vowel sounds are converging or diverging. In the U.S. South, the Southern vowel shift (SVS) and African American vowel shift (AAVS) affect not only vowels’ relative positions but also their formant dynamics. Static characterizations of shifting, with a single pair of first and second formant values taken near vowels’ midpoint, fail to capture this vowel-inherent spectral change, which can indicate dialect-specific diphthongization or monophthongization. Vowel-inherent spectral change is directly modeled to investigate how trajectories of front vowels /i eɪ ɪ ɛ/ differ across social groups in the 64-speaker Digital Archive of Southern Speech. Generalized additive mixed models are used to test the effects of two social factors, sex and ethnicity, on trajectory shape. All vowels studied show significant differences between men, women, African American and European American speakers. Results show strong overlap between the trajectories of /eɪ, ɛ/ particularly among European American women, consistent with the SVS, and greater vowel-inherent raising of /ɪ/ among African American speakers, indicating how that lax vowel is affected by the AAVS. Model predictions of duration additionally indicate that across groups, trajectories become more peripheral as vowel duration increases.Four self-identified code-switchers from Southwest Virginia and six actors who did not identify as having Southern accents each recorded two sets of stimuli in which they aimed to produce a more Southern and a more Standardized US accent. An analysis of the Voice Onset Time (VOT) of phrase and word initial voiced stops revealed that both groups of speakers produced more tokens with prevoicing (lead or negative lag voicing) when speaking in the Southern versus the Standard guise, and that in the Standard guise, the Southern speakers produced more prevoiced tokens than the actors. These findings support fairly recent descriptions of lead voicing as a feature of Southern US English. They additionally show that despite the lack of overt commentary about this feature, speakers have an awareness of the association between lead voicing and Southern US English because they manipulate the feature in a socially meaningful way; in Labov’s [(1972). Sociolinguistic Patterns (Blackwell, Oxford)] terminology, negative lag voicing is a marker of Southern US English.Quantitative ultrasound is used to characterize osseointegration at the bone-implant interface (BII). However, the interaction between an ultrasonic wave and the implant remains poorly understood. Hériveaux, Nguyen, and Haiat [(2018). J. Acoust. Soc. Am. 144, 488-499] recently employed a two-dimensional (2D) model of a rough BII to investigate the sensitivity of the ultrasonic response to osseointegration. The present letter aimed at assessing the validity of the 2D assumption. The values of the reflection coefficient of the BII obtained with two and three-dimensional models were found not to be significantly different for implant roughness lower than 20 μm. 2D modeling is sufficient to describe the interaction between ultrasound and the BII.A modal model for diffraction by a contiguous array of rectangular grooves in an acoustically-hard plane is extended to predict the free space acoustic field from a point source above such a structure. Subsequently, an approximate effective impedance model for grooved surfaces is presented. Measurements have shown that these ground surfaces can be used for outdoor noise reduction but accurate modelling has required the use of computationally expensive numerical methods. The extended modal model and approximate impedance model inspired by it yield equivalent results in a fraction of the time taken by the boundary element method, for example, and could be used when designing grooved surfaces to reduce noise from road traffic.This paper intends to explore the rationality and feasibility of modeling dispersed submicron particles in air by a kinetic-based method called the unified gas-kinetic scheme (UGKS) and apply it to the simulation of particle concentration under a transverse standing wave. Antineoplastic and I inhibitor A gas-particle coupling scheme is proposed where the gas phase is modeled by the two-dimensional linearized Euler equations (LEE) and, through the analogous behavior between the rarefied gas molecules and the air-suspended particles, a modified UGKS is adopted to estimate the particle dynamics. The Stokes’ drag force and the acoustic radiation force applied on particles are accounted for by introducing a velocity-dependent acceleration term in the UGKS formulation. To validate this methodology, the computed concentration patterns are compared with experimental results in the literature. The comparison shows that the adopted LEE-UGKS coupling scheme could well capture the concentration pattern of suspended submicron particles in a channel. In addition, numerical simulations with varying standing wave amplitudes, different acoustic radiation force to drag force ratios, and mean flow velocities are conducted. Their respective influences on the particle concentration pattern and efficiency are analyzed.In this work, a convolutional neural network (CNN) is applied to recognize acoustic spatial patterns with the aid of acoustic visualization. The acoustic spatial patterns are obtained by the singular value decomposition of an acoustic radiation operator built with the boundary integral equation. It is to explore the powerful capability of the CNN in the image processing by analogously rendering the measured acoustic spatial patterns into images. Due to practical limitations, a higher resolution of an acoustic image is achieved by interpolating the pressure on a coarse grid. Steady-state analysis of acoustic problems is a complex domain problem. The acoustic fields are then supplied into a CNN scheme as two-channel data which are real and imaginary components of the pressure. Random noises and incident waves with varying energy are added to the measured data to simulate influences from uncorrelated and correlated noises, respectively. It is demonstrated that once the CNN scheme is built and trained with adequate data, which is numerically synthesized, the patterns can be more accurately and robustly recognized by comparing it with the cross-correlation based methods.