Noise

About: Noise is a research topic. Over the lifetime, 5111 publications have been published within this topic receiving 69407 citations. The topic is also known as: Мопсы танцуют под радио бандитов из сталкера 10 часов.

System and method for improving audio speech quality

Abstract: The invention discloses a system and a method for improving audio speech quality comprises the following steps that: firstly, preemphasis and Fourier transform treatment of the current frame of time domain first audio signal output by an audio collecting device and the current frame of the time domain second audio signal output by an audio playing device are completed; secondly, according to the frequency domain error signal parameter and frequency domain learning rate parameter obtained by the previous frame of the current frame, a frequency domain second audio signal is corrected; thirdly, de-emphasis treatment and subtraction of the first audio signal and the second audio signal are completed, thereby eliminating the echo in the first audio signal; meanwhile, according to the correctedtime domain second audio signal and frequency domain second audio signal and preemphasized time domain first audio signal, the frequency domain error signal parameter and frequency domain learning rate parameter of the current frame are calculated and used for the treatment of the next frame, thereby effectively eliminating speech echo; in addition, through estimating the residual echo and noise of the audio signal with the echo eliminated, the invention effectively eliminates the noise in the audio signal and greatly improves the speech quality.

Speech enhancement based on a modified spectral subtraction method

Abstract: In this paper, a noisy speech enhancement method based on modified spectral subtraction performed on short time magnitude spectrum is presented. Here the cross-terms containing spectra of noise and clean signals are taken into consideration which are neglected in the traditional spectral subtraction method on the basis of the assumption that clean speech and noise signals are completely uncorrelated which is not true for most of the noises. In this method, the noise estimate to be subtracted from the noisy speech spectrum is proposed to be determined exploiting the low frequency regions of noisy speech of the current frame rather than depending only on the initial silence frames. We argue that this approach of noise estimation is capable of tracking the time variation of the non-stationary noise. By employing the noise estimates thus obtained, a procedure is formulated to reduce noise from the magnitude spectrum of noisy speech signal. The noise reduced magnitude spectrum is then recombined with the unchanged phase spectrum to produce a modified complex spectrum prior to synthesizing an enhanced frame. Extensive simulations are carried out using NOIZEUS database in order to evaluate the performance of the proposed method. It is shown in terms of objective measures, spectrogram analysis and subjective listening tests that the proposed method consistently outperforms one of the state-of-the-art methods of speech enhancement from noisy speech corrupted by babble or car noise at high as well as very low levels of SNR.

The effect of blindness on horizontal plane sound source identification

Abstract: The effect of blindness on sound source identification was studied. Four groups of normally-hearing adults, two sighted and two blind, participated. Subjects were tested using arrays of four and eight loudspeakers, surrounding them in the horizontal plane. One sighted group was tested in quiet. The other groups were tested in continuous 60-dB SPL white noise. Three 75-dB SPL 300-ms stimuli were localized: one-third octave noise bands, centered at 0.5 and 4 kHz, and broadband noise. Broadband noise was easiest to localize (both binaural and spectral cues available), and the 0.5-kHz noise band was the most difficult (primarily interaural temporal difference cue available). Subjects with late-onset blindness achieved significantly higher scores than the early blind and blindfolded sighted subjects. The percentage correct decreased with an increase in the number of speakers, but background noise had no effect. The results attest to the benefit of early visual experience for spatial hearing in adulthood, and demonstrate the negative impact of sudden loss of sight.

General-purpose audio tagging from noisy labels using convolutional neural networks

Abstract: General-purpose audio tagging refers to classifying sounds that are of a diverse nature, and is relevant in many applications where domain-specific information cannot be exploited. The DCASE 2018 challenge introduces Task 2 for this very problem. In this task, there are a large number of classes and the audio clips vary in duration. Moreover, a subset of the labels are noisy. In this paper, we propose a system to address these challenges. The basis of our system is an ensemble of convolutional neural networks trained on log-scaled mel spectrograms. We use preprocessing and data augmentation methods to improve the performance further. To reduce the effects of label noise, two techniques are proposed: loss function weighting and pseudo-labeling. Experiments on the private test set of this task show that our system achieves state-of-the-art performance with a mean average precision score of 0.951

Noise in disk data-recording media

Abstract: Measurements were made of recording medium noise in erased disks using an in-contact magnetoresistive element and an inductive head supported on an air bearing slider. Four types of coatings on aluminum disks were examined: thin, transition-metal alloy film, CrO,, FeCo particle, a n d y-Fe,O:,. Results obtained by means of three measurement techniques are in qualitative agreement and indicate that: (1) dc-erased noi\\e of alloy film disks is 14 to 20 dB lower than that of particulate disks measured; (2) dc-erased noise of particulate disks measured is 6 to 16 dB above their bulk-erased noise: (3) although dc noise of particulate disks increases with write current, dc noise of alloy film disks is independent of write current: (4) the shapes of the noise spectra are similar in dc-erased particulate y-Fe,O, disks and FeCo particle coated disks: and (5) significant modulation noise is detected on particulate disks but not on alloy film disks. The observed dc-erased noise spectrum is compared with the model for small particle noise and is then used to estimate the size of particle agglomerates or voids. Introduction Several theoretical and experimental investigations of noise in magnetic tape systems have been published [ 1-61, but until recently there was little interest in measuring the corresponding noise in disk surface media because it does not limit performance of present disk data storage systems. Disk noise measurements are of theoretical interest because of the relatively well-known head-medium spacing and the expected absence of noise caused by surface irregularities, tape flutter, and inverse magnetostrictive effects associated with head-tape contact. Although most tape noise studies have emphasized the noise of ac-erased media, the emphasis here is on the dc-erased, or uniformly magnetized, state which resembles the saturated digital state. The dc-erase noise provides information about the homogeneity of the magnetic medium, and its measurement can shed light on the suit-ability of the medium for use at high storage densities. The difference between the observed noise power spectrum and that which would be caused by small particles is interpreted as the noise spectrum of the inhomogenei-ties. Measurements of particulate coatings, both here and in the literature, indicate higher noise levels for dc erase than for ac erase. For comparison, measurements are presented here of a thin transition-metal alloy film medium, for which the reverse is true.