Showing papers on "Background noise published in 2002"

Speech pause detection for noise spectrum estimation by tracking power envelope dynamics

Abstract: A speech pause detection algorithm is an important and sensitive part of most single-microphone noise reduction schemes for enhancement of speech signals corrupted by additive noise as an estimate of the background noise is usually determined when speech is absent. An algorithm is proposed which detects speech pauses by adaptively tracking minima in a noisy signal's power envelope both for the broadband signal and for the high-pass and low-pass filtered signal. In poor signal-to-noise ratios (SNRs), the proposed algorithm maintains a low false-alarm rate in the detection of speech pauses while the standardized algorithm of ITU G.729 shows an increasing false-alarm rate in unfavorable situations. These characteristics are found with different types of noise and indicate that the proposed algorithm is better suited to be used for noise estimation in noise reduction algorithms, as speech deterioration may thus be kept at a low level. It is shown that in connection with the Ephraim-Malah (1984) noise reduction scheme, the speech pause detection performance can even be further increased by using the noise-reduced signal instead of the noisy signal as input for the speech pause decision unit.

Industrial Noise Control and Acoustics

Abstract: Introduction Basics of Acoustics Acoustic Measurements Transmission of Sound Noise Sources Acoustic Criteria Room Acoustics Silencer Design Vibration Isolation for Noise Control Case Studies in Noise Control Appendix A Preferred Prefixes in SI Appendix B Properties of Gases, Liquids, and Solids Appendix C Plate Properties of Solids Appendix D Surface Absorption Coefficients Appendix E Nomenclature Index

A psychoacoustic approach to combined acoustic echo cancellation and noise reduction

Abstract: This paper presents and compares algorithms for combined acoustic echo cancellation and noise reduction for hands-free telephones. A structure is proposed, consisting of a conventional acoustic echo canceler and a frequency domain postfilter in the sending path of the hands-free system. The postfilter applies the spectral weighting technique and attenuates both the background noise and the residual echo which remains after imperfect echo cancellation. Two weighting rules for the postfilter are discussed. The first is a conventional one, known from noise reduction, which is extended to attenuate residual echo as well as noise. The second is a psychoacoustically motivated weighting rule. Both rules are evaluated and compared by instrumental and auditive tests. They succeed about equally well in attenuating the noise and the residual echo. In listening tests, however, the psychoacoustically motivated weighting rule is mostly preferred since it leads to more natural near end speech and to less annoying residual noise.

Noise adaptive stream weighting in audio-visual speech recognition

Abstract: It has been shown that integration of acoustic and visual information especially in noisy conditions yields improved speech recognition results. This raises the question of how to weight the two modalities in different noise conditions. Throughout this paper we develop a weighting process adaptive to various background noise situations. In the presented recognition system, audio and video data are combined following a Separate Integration (SI) architecture. A hybrid Artificial Neural Network/Hidden Markov Model (ANN/HMM) system is used for the experiments. The neural networks were in all cases trained on clean data. Firstly, we evaluate the performance of different weighting schemes in a manually controlled recognition task with different types of noise. Next, we compare different criteria to estimate the reliability of the audio stream. Based on this, a mapping between the measurements and the free parameter of the fusion process is derived and its applicability is demonstrated. Finally, the possibilities and limitations of adaptive weighting are compared and discussed.

Detecting voiced and unvoiced speech using both acoustic and nonacoustic sensors

Abstract: Systems and methods are provided for detecting voiced and unvoiced speech in acoustic signals having varying levels of background noise. The systems receive acoustic signals at two microphones, and generate difference parameters between the acoustic signals received at each of the two microphones. The difference parameters are representative of the relative difference in signal gain between portions of the received acoustic signals. The systems identify information of the acoustic signals as unvoiced speech when the difference parameters exceed a first threshold, and identify information of the acoustic signals as voiced speech when the difference parameters exceed a second threshold. Further, embodiments of the systems include non-acoustic sensors that receive physiological information to aid in identifying voiced speech.

Classification and characterization of impulsive noise on indoor powerline used for data communications

Abstract: The performance of a link using the indoor power line network as a medium for communication strongly depends on the noise characteristics. Besides the background noise and the narrow band noise mainly due to broadcast transmitters, impulsive noise adversely affects the quality of service. This paper focuses on impulsive noise measurements, their statistical properties being the basis of a noise model for optimizing a transmission scheme.

Speech enhancement using a mixture-maximum model

Abstract: We present a spectral domain, speech enhancement algorithm. The new algorithm is based on a mixture model for the short time spectrum of the clean speech signal, and on a maximum assumption in the production of the noisy speech spectrum. In the past this model was used in the context of noise robust speech recognition. In this paper we show that this model is also effective for improving the quality of speech signals corrupted by additive noise. The computational requirements of the algorithm can be significantly reduced, essentially without paying performance penalties, by incorporating a dual codebook scheme with tied variances. Experiments, using recorded speech signals and actual noise sources, show that in spite of its low computational requirements, the algorithm shows improved performance compared to alternative speech enhancement algorithms.

Estimating stationary dipoles from MEG/EEG data contaminated with spatially and temporally correlated background noise

Abstract: The stationary dipole model for the inverse problem of magnetoencephalographic (MEG) and electroencephalographic (EEG) data is extended by including spatio-temporal correlations of the background noise. For that purpose, the spatio-temporal covariances are described as a Kronkecker product of a spatial and a temporal covariance matrix. The maximum likelihood method is used to estimate this Kronecker product from a series of trials of MEG/EEG data. A simulation study shows that the inclusion of the background noise generally improves the dipole estimate substantially. When the frequency of the source time functions, however, coincides with the frequency contents of the covariance function, the dipole estimate worsens when the temporal correlations are included. The inclusion of spatial correlations always improves the estimates.

An integrated audio and active noise control headset

Abstract: This paper presents an integrated approach in designing a noise reduction headset for audio and communication applications. Conventional passive headsets give good attenuation of ambient noise in the upper frequency range, while most of these devices fail below 500 Hz. Unlike the feedforward method, the adaptive feedback active noise control technique provides more accurate noise cancellation since the microphone is placed inside the ear-cup of the headset. Furthermore, the system uses a single microphone per ear cup, thus producing a more compact, lower power consumption, cheaper solution and ease of integration with existing audio and communication devices to form an integrated feedback active noise control headset. Simulation results have been conducted to show that the integrated approach can remove the disturbing noise and, at the same time, allow the desired speech or audio signal to pass through without cancellation.

Effect of noise and occupancy on optimal reverberation times for speech intelligibility in classrooms.

Abstract: The question of what is the optimal reverberation time for speech intelligibility in an occupied classroom has been studied recently in two different ways, with contradictory results. Experiments have been performed under various conditions of speech-signal to background-noise level difference and reverberation time, finding an optimal reverberation time of zero. Theoretical predictions of appropriate speech-intelligibility metrics, based on diffuse-field theory, found nonzero optimal reverberation times. These two contradictory results are explained by the different ways in which the two methods account for background noise, both of which are unrealistic. To obtain more realistic and accurate predictions, noise sources inside the classroom are considered. A more realistic treatment of noise is incorporated into diffuse-field theory by considering both speech and noise sources and the effects of reverberation on their steady-state levels. The model shows that the optimal reverberation time is zero when the speech source is closer to the listener than the noise source, and nonzero when the noise source is closer than the speech source. Diffuse-field theory is used to determine optimal reverberation times in unoccupied classrooms given optimal values for the occupied classroom. Resulting times can be as high as several seconds in large classrooms; in some cases, optimal values are unachievable, because the occupants contribute too much absorption.

Vocal loading among day care center teachers.

Abstract: Day care center teachers suffer from voice disorders more often than nurses do. Several risk factors may increase voice disorder prevalence of day care center teachers. The risk factors can be bound to their job content and manner of working i.e. having to raise their voice to attract the attention of the children and to offer them the possibility to perceive spoken information, or to the environment i.e. poor acoustics and excess background noise. The purpose of this study was to measure some of the risk factors for voice disorders of day care center teachers and of a control group (nurses); these were speaking times and speech levels. The background noise levels during activities and RASTI-values (Rapid Speech Transmission Index), i.e. measures of the acoustics of rooms, were also measured at the day care centers. It was found that day care center teachers use their voices more and with higher levels than nurses do. It was also found that the background noise levels are high, which is partly due to the ...

SQUID sensor array configurations for magnetoencephalography applications

Abstract: Electrophysiological activity in the human brain generates a small magnetic field from the spatial superposition of individual neuronal source currents. At a distance of about 15 mm from the scalp, the observed field is of the order of 10−13 to 10−12 T peak-to-peak. This measurement process is termed magnetoencephalography (MEG). In order to minimize instrumental noise, the MEG is usually detected using superconducting flux transformers, coupled to SQUID (superconducting quantum interference device) sensors. Since MEG signals are also measured in the presence of significant environmental magnetic noise, flux transformers must be designed to strongly attenuate environmental noise, maintain low instrumental noise and maximize signals from the brain. Furthermore, the flux transformers must adequately sample spatial field variations if the brain activity is to be imaged. The flux transformer optimization for maximum brain signal-to-noise ratio (SNR) requires analysis of the spatial and temporal properties of brain activity, the environmental noise and how these signals are coupled to the flux transformer. Flux transformers that maximize SNR can detect the smallest brain signals and have the best ability to spatially separate dipolar sources. An optimal flux transformer design is a synthetic higher-order gradiometer based on relatively short-baseline first-order radial gradiometer primary sensors.

Broadband Seismic Background Noise at Temporary Seismic Stations Observed on a Regional Scale in the Southwestern United States

Abstract: Background noise power spectral density (PSD) estimates for 54 PASS- CAL Colorado Plateau/Rio Grande Rift/Great Plains Seismic Transect (LA RISTRA) stations were computed using data from 1999 to 2000. At long periods (0.01-0.1 Hz), typical vertical noise levels are approximately 12 dB higher than the nearby Global Seismic Network (GSN) borehole station ANMO, but horizontal power spec- tral density (PSD) noise levels are approximately 30 dB higher. Long-period noise levels exhibit essentially no spatial correlation along the LA RISTRA transect, indi- cating that local thermal or atmosphere-driven local slab tilt is the dominant source of noise in this band. Between 0.1 and 0.3 Hz, typical noise levels are dominated by naturally occurring microseismic noise and are essentially identical to those observed at ANMO. At short periods, 0.3-8 Hz, typical noise levels along the network exceed ANMO levels by approximately 15 dB, with the highest levels corresponding to proximity to cultural noise sources. No significant day/night variations were observed in the microseismic band; however, both low- and high-frequency noise levels show an increase of up to 8 dB in median midday versus midnight noise levels. We find that the major shortcomings of these shallow PASSCAL-style temporary vaults rela- tive to a GSN-style borehole installation are increased susceptibility to long-period horizontal (20 sec) noise and to surface noise sources above approximately 2 Hz. Although the high-frequency near-surface noise field is unavoidable in shallow vaults, we suggest that increased understanding and mitigation of local tilt effects in shallow vaults offers the possibility of significantly improving the long-period noise environment.

Measurement of 1/f noise and its application in materials science

Abstract: This is a review of the measurement of 1/f noise in certain classes of materials which have a wide range of potential applications. This includes metal films, semiconductors, metallic oxides and inhomogeneous systems such as composites. The review contains a basic introduction to this field, the theories and models and follows it up with a discussion on measurement methods. There are discussions on specific examples of the application of noise spectroscopy in the field of materials science.

Adaptive filtering of evoked potentials with radial-basis-function neural network prefilter

Abstract: Evoked potentials (EPs) are time-varying signals typically buried in relatively large background noise. To extract the EP more effectively from noise, we had previously developed an approach using an adaptive signal enhancer (ASE) (Chen et al., 1995). ASE requires a proper reference input signal for its optimal performance. Ensemble- and moving window-averages were formerly used with good results. In this paper, we present a new method to provide even more effective reference inputs for the ASE. Specifically, a Gaussian radial basis function neural network (RBFNN) was used to preprocess raw EP signals before serving as the reference input. Since the RBFNN has built-in nonlinear activation functions that enable it to closely fit any function mapping, the output of RBFNN can effectively track the signal variations of EP. Results confirmed the superior performance of ASE with RBFNN over the previous method.

Weighted median image sharpeners for the World Wide Web

Abstract: A class of robust weighted median (WM) sharpening algorithms is developed in this paper. Unlike traditional linear sharpening methods, weighted median sharpeners are shown to be less sensitive to background random noise or to image artifacts introduced by JPEG and other compression algorithms. These concepts are extended to include data dependent weights under the framework of permutation weighted medians leading to tunable sharpeners that, in essence, are insensitive to noise and compression artifacts. Permutation WM sharpeners are subsequently generalized to smoother/sharpener structures that can sharpen edges and image details while simultaneously filter out background random noise. A statistical analysis of the various algorithms is presented, theoretically validating the characteristics of the proposed sharpening structures. A number of experiments are shown for the sharpening of JPEG compressed images and sharpening of images with background film-grain noise. These algorithms can prove useful in the enhancement of compressed or noisy images posted on the World Wide Web (WWW) as well as in other applications where the underlying images are unavoidably acquired with noise.

Solar p-modes from 1979 days of the GOLF experiment

Abstract: With the GOLF instrument onboard the SoHO observatory, 1979 days of full-disc Doppler velocity observations have been compiled into a study of p-mode properties. We develop a multi-step iterative method (MSIM) algorithm to access all p-mode parameters while minimizing any perturbating effect or cross-talk between parameters during their determination. We present frequency and splitting tables, amplitudes, linewidths, line asymmetries, pseudo-modes, and background noise determinations. We have a first look at the changes induced by the transition from the low-activity to the high-activity part of solar cycle 23: we have recorded frequency shifts with a downturn at 3.7 mHz followed by a possible higher upturn, and linewidth changes to a good accuracy. We detect an effect on the noise background at 3 mHz possibly related to an interaction between noise and the modes and connected to the asymmetry of the profiles.

Deconvolution improves colocalization analysis of multiple fluorochromes in 3D confocal data sets more than filtering techniques

Abstract: Background and noise impair image quality by affecting resolution and obscuring image detail in the low intensity range. Because background levels in unprocessed confocal images are frequently at about 30% maximum intensity, colocalization analysis, a typical segmentation process, is limited to high intensity signal and prone to noise-induced, false-positive events. This makes suppression or removal of background crucial for this kind of image analysis. This paper examines the effects of median filtering and deconvolution, two image-processing techniques enhancing the signal-to-noise ratio (SNR), on the results of colocalization analysis in confocal data sets of biological specimens. The data show that median filtering can improve the SNR by a factor of 2. The technique eliminates noise-induced colocalization events successfully. However, because filtering recovers voxel values from the local neighbourhood false-negative ('dissipation' of signal intensity below threshold value) as well as false-positive ('fusion' of noise with low intensity signal resulting in above threshold intensities), results can be generated. In addition, filtering involves the convolution of an image with a kernel, a procedure that inherently impairs resolution. Image restoration by deconvolution avoids both of these disadvantages. Such routines calculate a model of the object considering various parameters that impair image formation and are able to suppress background down to very low levels (< 10% maximum intensity, resulting in a SNR improved by a factor 3 as compared to raw images). This makes additional objects in the low intensity but high frequency range available to analysis. In addition, removal of noise and distortions induced by the optical system results in improved resolution, which is of critical importance in cases involving objects of near resolution size. The technique is, however, sensitive to overestimation of the background level. In conclusion, colocalization analysis will be improved by deconvolution more than by filtering. This applies especially to specimens characterized by small object size and/or low intensities.

Detecting small moving objects using temporal hypothesis testing

Abstract: This paper addresses the problem of detecting small, moving, low amplitude objects in image sequences that also contain moving nuisance objects and background noise. We formulate this problem in the context of a hypothesis testing procedure on individual pixel temporal profiles, leading to a computationally efficient statistical test. The technique assumes we have reasonable deterministic and statistical models for the temporal behavior of the background noise, target, and clutter, on a single pixel basis. Based on these models we develop a generalized likelihood ratio test (GLRT) and perfect measurement performance analysis, and present the resulting decision rule. We also propose a parameter estimation technique and compare its performance to the Cramer Rao bound (CRB). We demonstrate the effectiveness of the technique by applying the resulting algorithm to real world infrared (IR) image sequences containing targets of opportunity. The approach could also be applicable to other image sequence processing scenarios, using acquisition systems besides IR imaging, such as detection of small moving objects or structures in a biomedical or biological imaging scenario, or the detection of satellites, meteors or other celestial bodies in night sky imagery acquired using a telescope.

Children's use of semantic cues in degraded listening environments.

Abstract: Children 5 and 9 years of age and adults were required to identify the final words of low- and high-context sentences in background noise. Age-related differences in the audibility of speech signals were minimized by selecting signal-to-noise ratios (SNRs) that yielded 78% correct performance for low-context sentences. As expected, children required more favorable SNRs than adults to achieve comparable levels of performance. A more difficult listening condition was generated by adding 2 dB of noise. In general, 5-year-olds performed more poorly than did 9-year-olds and adults. Listeners of all ages, however, showed comparable gains from context in both levels of noise, indicating that noise does not impede children’s use of contextual cues.

Advanced interferometric phase and amplitude noise measurements

Abstract: The measurement of the close-to-the-carrier noise of two-port radio frequency and microwave devices is a relevant issue in time and frequency metrology and in some fields of electronics, physics, and optics. While phase noise is the main concern, amplitude noise is often of interest. Presently the highest sensitivity is achieved with the interferometric method, that consists of the amplification and synchronous detection of the noise sidebands after suppressing the carrier by vector subtraction of an equal signal. A substantial progress in understanding the flicker noise mechanism of the interferometer results in new schemes that improve by 20–30 dB the sensitivity at low Fourier frequencies. These schemes, based on two or three nested interferometers and vector detection of noise, also feature closed-loop carrier suppression control, simplified calibration, and intrinsically high immunity to mechanical vibrations. This article provides the complete theory and detailed design criteria, and reports on the implementation of a prototype working at the carrier frequency of 100 MHz. In real-time measurements, a background noise of −175 to −180 dBrad2/Hz has been obtained at f=1 Hz off the carrier; the white noise floor is limited by the thermal energy kBT0 referred to the carrier power P0 and by the noise figure of an amplifier. Exploiting correlation and averaging in similar conditions, the sensitivity exceeds −185 dBrad2/Hz at f=1 Hz; the white noise floor is limited by thermal uniformity rather than by the absolute temperature. A residual noise of −203 dBrad2/Hz at f=250 Hz off the carrier has been obtained, while the ultimate noise floor is still limited by the averaging capability of the correlator. This is equivalent to a signal-to-noise (S/N) ratio of 2×1020 with a frequency spacing of 2.5×10−6. All these results have been obtained in a relatively unclean electromagnetic environment, and without using a shielded chamber. Implementation and experiments at that sensitivity level require skill and tricks, for which a great effort is spent in the article. Applications include the measurement of the properties of materials and the observation of weak flicker-type physical phenomena, out of reach for other instruments. As an example, we measured the flicker noise of a by-step attenuator (−171 dBrad2/Hz at f=1 Hz) and of the ferrite noise of a reactive power divider (−173.7 dBrad2/Hz at f=1 Hz) without need of correlation. In addition, the real-time measurements can be exploited for the dynamical noise correction of ultrastable oscillators.

Simulation of cDNA microarrays via a parameterized random signal model.

Abstract: cDNA microarrays provide simultaneous expression mea- surements for thousands of genes that are the result of processing images to recover the average signal intensity from a spot composed of pixels covering the area upon which the cDNA detector has been put down. The accuracy of the signal measurement depends on using an appropriate algorithm to process the images. This includes deter- mining spot locations and processing the data in such a way as to take into account spot geometry, background noise, and various kinds of noise that degrade the signal. This paper presents a stochastic model for microarray images. There are over 20 model parameters, each governed by a probability distribution, that control the signal intensity, spot geometry, spot drift, background effects, and the many kinds of noise that affect microarray images owing to the manner in which they are formed. The model can be used to analyze the performance of image algorithms designed to measure the true signal intensity be- cause the ground truth (signal intensity) for each spot is known. The levels of foreground noise, background noise, and spot distortion can be set, and algorithms can be evaluated under varying conditions.

A robust endpoint detection of speech for noisy environments with application to automatic speech recognition

Abstract: We propose a new approach for classifying speech vs. non-speech, which proves to significantly improve speech recognition performance under noise. The proposed algorithm relies on the energy and spectral characteristics of the signal and applies a 3-level two-dimensional thresholding to determine whether an input frame is speech or non-speech. The algorithm runs in real-time, and offers better immunity to background noise, and to background speech than traditional energy-based word boundary detection. The performance of the endpoint detector is reported here in terms of improvements in speaker-independent (SI) and speaker-dependent (SD) recognition performance using 5 different simulated noise conditions and various signal-to-noise ratios (SNR). The proposed endpoint detection of speech improves the SD recognition accuracy by 24% for office noise, and reduces the false rejection rates for both SI and SD by 45% for babble noise and lobby noise.

Neural action potential detector using multi-resolution TEO

Abstract: Detection of action potentials in background noise is the first step to many neural researches including neural prostheses. The detectors using a Teager energy operator (TEO) are very simple and efficient, but sensitive to the background noise spectrum. A new detector using a multi-resolution TEO (MTEO) provides considerable improvement.