
This study presents a detailed acoustic analysis of the vowel inventory of Brajbhasha, a central Indo-Aryan language, examining its ten oral monophthongs. Through formant and duration measurements, the analysis evaluates the system's structure against the predictions of Dispersion Theory vs Theory of Adaptive Dispersion, which predict “maximal” vs “sufficient” dispersion (respectively) in the vowel space. Results reveal that vowels are acoustically distinct, but their dispersion in the F1-F2 plane does not achieve maximality, demonstrated through the simulation of 10 000 dispersions (using a novel empirical convex hull constraint). Instead, contrasts are maintained through a strategic spectral complementarity: vowel pairs that overlap in one spectral dimension (height or backness) are well-separated in the other. Furthermore, duration exists as a critical, phonetic cue: three inner vowels /ɪ, ɐ, ʊ/ are approximately 65 ms shorter (metrically half) than the other seven. Including duration in linear discriminant analysis significantly improves vowel classification accuracy from 83.5% to 88.1%. The findings characterize Brajbhasha's vowel inventory as a mixed-quantity system, with perceptual distinctiveness achieved through a strategic spectral-temporal complementarity, rather than maximal dispersion alone. This acoustic baseline also provides groundwork for lexical tonal research in the language.
Active noise control (ANC) is an effective approach to attenuate unwanted sound, but conventional adaptive algorithms, such as the filtered-x least mean square (FxLMS), are limited in the presence of fast-transient disturbances and exhibit a spatially restricted quiet zone. This work presents a novel model-based ANC strategy that addresses these limitations by combining virtual sensing, future disturbance prediction, and secondary path inversion within a unified state-space framework. Kalman filter-based virtual sensing enables the relocation of the quiet zone to a desired target location without additional physical sensors, while a dead time system predictor anticipates fast-varying disturbances to overcome the convergence delay of adaptive filters. The predicted disturbance is then processed through a minimum-phase inverse of the secondary path to generate the control signal in advance, ensuring accurate cancellation at the target position. The proposed methodology is validated through numerical simulations and experimental tests on a Y-shaped Kundt's tube representing a ducted heating, ventilation, and air conditioning (HVAC) system. Results demonstrate effective attenuation of broadband and fast-transient noise at the virtual microphone location, with additional beneficial effects observed outside the controlled point. The approach provides a robust framework for practical ANC applications in environments where rapid noise variations and constrained sensor placement challenge conventional control strategies.
Handheld hydraulic tools, including hammer drills and chipping hammers, are used in offshore construction. A hydrophone was used to measure underwater sound radiated by drilling and chipping hammers during anchor maintenance activities in Kaneohe Bay, Hawai'i. The recordings were used to calculate mean-square and zero-to-peak sound pressure levels, sound exposure levels, and to estimate acoustic source levels. Both tools had estimated source levels between 158 and 177 dB re 1 μPa-m with peak sound pressures of approximately 1.6 kPa at 10 m during the most intense periods of operation. The sound produced by both tools is highly percussive and intermittent with peak intensities in the 1–10 kHz band. Cumulative, unweighted sound exposure levels associated with both activities range from approximately 150 dB re 1 μPa2-s after 5 min to 160 dB re 1 μPa2-s or more after 40 min at a range of approximately 90 m. Although the most intense sound corresponding to the drilling and chipping activities results in modest (<8 dB) decreases in sound exposure levels for high-frequency hearing specialists, this same group is most likely to experience temporary or permanent threshold shifts because of these activities given their lower onset thresholds.
Harbour porpoise (Phocoena phocoena) densities are estimated routinely from acoustic detections. Previous acoustic studies show the probability of detecting porpoises (or conversely false negative rate) depends strongly on porpoise behaviour, changing density estimates several-fold if unaccounted for. We quantify this in widely used automated click detectors (e.g., C-PODs), where post-processing is used to identify sequences of consecutive clicks (“click trains”) and summarise results as detection-positive time units (e.g., click-positive seconds or minutes). We explored how post-processing choices change detectability using 228 h of fine-scale acoustic and movement data from tags on ten harbour porpoises in Danish waters. We simulated the click detection process, given assumed porpoise beam pattern and sound propagation, ran a custom click-train classifier on the resulting data and summarized results over different time units. Using porpoise-positive minutes after click-train classification as a metric, the most detectable behaviour (pelagic feeding) averaged 2.36 times [95% confidence interval 1.81–3.87] as detectable as the least (non-feeding) due to high click and turning rates. Click-train classification increased detectability differences between pelagic feeding and other behaviours. PAM studies may appreciably over- or under-estimate animal density; click-train classification may magnify this effect. We provide guidance on how these issues could be quantified and mitigated.
This work develops an analytic model for acoustic diffraction by finite barriers of arbitrary impedance from an extended, partially coherent noise source. The barrier impedance is incorporated through transmission and reflection coefficients obtained from standard layer analysis, while finite-barrier effects are modeled by extending analytic diffraction solutions to allow for multiple edges. The formulation is expressed in a transfer-matrix framework that maps source strengths to the radiated pressure field and is generalized to partially coherent sources using cross-spectral matrices. This framework is applied to an example problem that explores the potential of a curtain of bubbly water as a jet noise reduction mechanism for nearby launch personnel. While such a curtain can reduce noise significantly, the acoustic impact of bubbles is negligible under realistic conditions. A best-case scenario is also examined by treating the barrier as an anechoic surface. Under this ideal scenario, it is shown that noise reduction is fundamentally limited by diffraction. Finally, the importance of including coherence in the source description is discussed in the context of the radiated field.