Simulated Background Noise Research Articles

Autoencoder-extended Conditional Invertible Neural Networks for Unfolding Signal Traces

The reconstruction of cosmic ray-induced air showers from measurements of radio waves constitutes a major challenge. In this work, we focus on recovering the full three-dimensional electromagnetic field from two recorded signal traces of an antenna station covering two horizontal polarization directions. The simulated field is folded by a direction and frequency-dependent characteristic antenna response pattern, resulting in voltage signal traces as a function of time. Both signal traces are contaminated by simulated background noise. We use conditional Invertible Neural Networks (cINNs) to learn posterior distributions, from which the most likely electromagnetic field given a measured signal trace can be inferred. To improve robustness, we extend the method with an autoencoder by reducing the parameter phase space and decoupling the cINN from specific data shapes. Thereby, each signal trace is condensed into a small number of abstract parameters in the latent space on which the cINN operates. The presented method shows promising results and can be transferred to other unfolding problems where the recovery of the pre-measurement state is of interest.

The negative impact of wearing personal protective equipment on communication during coronavirus disease 2019.

Coronavirus disease 2019 personal protective equipment has been reported to affect communication in healthcare settings. This study sought to identify those challenges experimentally. Bamford-Kowal-Bench speech discrimination in noise performance of healthcare workers was tested under simulated background noise conditions from a variety of hospital environments. Candidates were assessed for ability to interpret speech with and without personal protective equipment, with both normal speech and raised voice. There was a significant difference in speech discrimination scores between normal and personal protective equipment wearing subjects in operating theatre simulated background noise levels (70 dB). Wearing personal protective equipment can impact communication in healthcare environments. Efforts should be made to remind staff about this burden and to seek alternative communication paradigms, particularly in operating theatre environments.

An automated assessment system of limits of detection and quantitation in gradient high-performance liquid chromatography with ultraviolet detection

A previous paper of this series of study put forward a basic model of an automated system for predicting detection limits and showed its application to a simple example of isocratic high-performance liquid chromatography (HPLC). This paper describes an expansion of the basic system into gradient HPLC. The most serious problem with the expansion is a long-term variation in backgrounds, called gradient baseline drifts, which in theory cannot be covered by a noise model (stationary random process) of the original system. This paper demonstrates that the above problem can be solved with modifying a parametrization procedure of the noise model. The essential role of the system is to predict the standard deviation (SD) of measurements at low concentrations from a chromatogram without repeated measurements of real samples. Laboratory-made software enables the automated assessment of the limits of detection and quantitation for each of chromatographically separated signals in a single run. Simulated background noise which consists of the stationary noise model with linear slopes is used to confirm the accuracy and reproducibility of the automated prediction. A gradient HPLC determination for cefaclor is taken as an example. The parametrization modification improves the correlation coefficient, r2, between the observed and theoretical distributions of the area measurements from 0.373 to 0.966. The statistical confidence levels of the theoretically predicted relative SDs for cefaclor were verified by comparing them with those obtained by repeated experiments (r2 = 0.989). The limits of detection (= 3.3 × SD = 18.0 µg/L) and quantitation (= 10 × SD = 54.7 µg/L) for cefaclor have signal-to-noise ratios close to the commonly adopted values, 3 and 10, respectively.

Audio-visual word prominence detection from clean and noisy speech

In this paper we investigate the audio-visual processing of linguistic prosody, more precisely the detection of word prominence, and examine how the additional visual information can be used to increase the robustness when acoustic background noise is present. We evaluate the detection performance for each modality individually and perform experiments using feature and decision fusion. For the latter we also consider the adaptive fusion with fusion weights adjusted to the current acoustic noise level. Our experiments are based on a corpus with 11 English speakers which contains in addition to the speech signal also videos of the speakers’ heads. From the acoustic signal we extract features which are well known to capture word prominence like loudness, fundamental frequency and durational features. The analysis of the visual signal is based on features derived from the speaker’s rigid head movements and movements of the speaker’s mouth. We capture the rigid head movements by tracking the speaker’s nose. Via a two-dimensional Discrete Cosine Transform (DCT) calculated from the mouth region we represent the movements of the speaker’s mouth. The results show that the rigid head movements as well as movements inside the mouth region can be used to discriminate prominent from non-prominent words. The audio-only detection yields an Equal Error Rate (EER) averaged over all speakers of 13%. Based only on the visual features we obtain 20% of EER. When we combine the visual and the acoustic features we only see a small improvement compared to the audio-only detection for clean speech. To simulate background noise we added 4 different noise types at varying SNR levels to the acoustic stream. The results indicate that word prominence detection is quite robust against additional background noise. Even at a severe Signal to Noise Ratio (SNR) of −10 dB the EER only rises to 35%. Despite this the audio-visual fusion leads to notable improvements for the detection from noisy speech. We observe relative reductions of the EER of up to 79%.

Development of crescentic bars for a periodically perturbed initial bathymetry

[1] The development of crescentic bed patterns starting from an ‘undisturbed’ beach (i.e. one only perturbed so as to simulate background noise and thereby initiate morphological development) is compared with that starting from a beach where monochromatic (and in some cases bichromatic) bed patterns pre-exist, using a fully non-linear model (Morfo55). A wide range of different lengthscales and amplitudes is investigated. The findings suggest that the pre-existence of crescentic bed-forms can influence the subsequent morphological development of this beach significantly, but that this development is related to the undisturbed beach development. Whether a pre-existing bed-form remains under certain forcing conditions depends on the position of the pre-existing lengthscale along the undisturbed linear growth rate curve: If the pre-existing mode shows significant linear growth in the undisturbed scenario, this pre-existing lengthscale remains. Otherwise the pre-existing mode will be replaced by a bed-form with a lengthscale closer to the undisturbed fastest growing mode. A pre-existing lengthscale that is significantly longer than the undisturbed fastest growing mode will be replaced by a higher harmonic of the original lengthscale; for shorter pre-existing lengthscales, this will be the linear fastest growing mode. An increased amplitude of the pre-existing bed-form accelerates the development toward a new stable situation. For small initial amplitudes, the initial response of the system to pre-existing bed-forms could theoretically be described using linearized equations alone. The migration rates of both the pre-existing and newly arising modes correspond to the migration rate of these modes in the undisturbed scenario.

Simulating the effect binaural (electric and acoustic) hearing on Mandarin speech recognition under background noise for patients with cochlear implant (CI)

This study evaluated the effects of binaurally combined electric and acoustic hearing on Mandarin speech recognition under simulated background noise for normal‐hearing (NH) listeners and CI recipients of the Nucleus‐24 device with contralateral hearing aid (HA). Chinese word and tone recognition were measured in seven NH subjects using simulated four‐channel continuous interleaved sampler (CIS) type speech processing strategy under three different conditions, namely, (1) only simulated (CI) signals with modulated Gaussian white noise for monaural stimulation, (2) simulated (CI) signals with modulated Gaussian white noise and simulated HA signals (low‐pass speech signals with cutoff frequency of 500 Hz) for monaural stimulation, and (3) simulated (CI) signals with modulated Gaussian white noise to one ear and simulated hearing aid (HA) signals to the other ear for dichotic stimulation. For comparison purposes, Chinese word and tone recognition were also measured in three CI users using advanced combination encoding (ACE) strategy. Preliminary results showed that the Chinese word and tone recognition rates of the NH listeners and CI recipients are statistically significantly improved (word and tone recognition rates were improved from 3.85% to 60.86% and 40.14% to 97.43% for NH listeners and 20% to 37.5% and 50% to 90% for CI recipients).

Relating human productivity and annoyance to indoor noise criteria systems

The goal of this research is to determine a noise criteria system which best relates the effects of background noise to human productivity and annoyance. A number of indoor noise criteria systems are currently used to rate the background noise in built environments, including noise criteria (NC), balanced noise criteria (NCB), room criteria (RC), room criteria Mark II (RC-Mark II), and others. Many questions still remain about the accuracy of these predictors in assessing human response to background noise under the variety of ambient noise situations encountered. To support the use of any individual criterion, subjective testing was performed under a range of background noise situations and statistically related to the various noise criteria predictors listed above. Subjects completed an annoyance survey and performed typing and proofreading tasks in an acoustically controlled environment under 12 simulated background noise settings. These settings varied across three sound levels and four spectral qualities. Subjective testing methodology and results are presented. [Work supported by INCE and ASHRAE.]

Determination of sound pressure levels in situ using sound intensity measurements

Measurements of emission sound pressure levels of machinery require either specially defined test rooms or calculated corrections for the acoustic environment. In principle, it is possible to determine emission sound pressure levels from sound intensity measurements at specified work stations in any test environment if the requirements of background noise levels and field indicators are fulfilled. The draft international standard ISO 11205/CD specifies such a method. In this paper the accuracy of emission sound pressure levels using sound intensity measurements was examined for three small sources in three acoustic environments, an anechoic environment with loudspeakers to simulate background noise, an office environment, and a reverberant environment inside a stairwell. In the first two environments good measurement accuracies, within 1 dB, were obtained. Sound intensity measurements by pointing the probe towards the source were as accurate, and simpler than computation of the resultant intensity using three arbitrary orthogonal measurements. As predicted by field indicators, measurements in the stairwell gave unacceptable errors for all three sources.

Digital filter design considerations for sonar trainer noise generation

Digital filters were designed and built to realistically simulate background noise for a submarine sonar operational training system. Broadband spectral models of rain and distant shipping were implemented using shift register noise generators followed by digital filters. Marine-life noise was reconstructed by playing actual tape recordings through a digital filter which accounted for propagation characteristics. Decimators and interpolators were used to alter sample rates for the various sonar channels. Spatial effects of noise were dynamically controlled with inverse beamformers. The effect of word size, bias, aliasing, and spatial control will be discussed. Also, audio examples of the simulated background noise will be presented. [Work supported by NUSC, New London.]

Simulated Background Noise to Improve Acoustical Privacy in Buildings

In many existing buildings where the construction provides a low degree of sound isolation between rooms, complaints concerning a lack of acoustical privacy most frequently arise where the ambient-background-noise levels are low. In most cases, the adequate improvement of the construction is either impractical or too costly. An increase of the ambient-background-noise level, on the other hand, can often provide the desired improvement of privacy. Sometimes this can be achieved by modifying an existing ventilating system. Where this cannot be done, background noise duplicating the characteristics of ventilation-system diffusers and other types of air distribution devices can be simulated electronically and distributed by means of loudspeakers. The instrumentation and technique as used in a few recent building installations is described.