Dataset Information

Segregating complex sound sources through temporal coherence.

ABSTRACT: A new approach for the segregation of monaural sound mixtures is presented based on the principle of temporal coherence and using auditory cortical representations. Temporal coherence is the notion that perceived sources emit coherently modulated features that evoke highly-coincident neural response patterns. By clustering the feature channels with coincident responses and reconstructing their input, one may segregate the underlying source from the simultaneously interfering signals that are uncorrelated with it. The proposed algorithm requires no prior information or training on the sources. It can, however, gracefully incorporate cognitive functions and influences such as memories of a target source or attention to a specific set of its attributes so as to segregate it from its background. Aside from its unusual structure and computational innovations, the proposed model provides testable hypotheses of the physiological mechanisms of this ubiquitous and remarkable perceptual ability, and of its psychophysical manifestations in navigating complex sensory environments.

SUBMITTER: Krishnan L

PROVIDER: S-EPMC4270434 | biostudies-literature | 2014 Dec

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Segregating complex sound sources through temporal coherence.

Krishnan Lakshmi L Elhilali Mounya M Shamma Shihab S

PLoS computational biology 20141218 12

A new approach for the segregation of monaural sound mixtures is presented based on the principle of temporal coherence and using auditory cortical representations. Temporal coherence is the notion that perceived sources emit coherently modulated features that evoke highly-coincident neural response patterns. By clustering the feature channels with coincident responses and reconstructing their input, one may segregate the underlying source from the simultaneously interfering signals that are unc ...[more]

PMID: 25521593

Similar Datasets

Project description:Previous studies on the effect of spectral content on auditory distance perception (ADP) focused on the physically measurable cues occurring either in the near field (low-pass filtering due to head diffraction) or when the sound travels distances >15 m (high-frequency energy losses due to air absorption). Here, we study how the spectrum of a sound arriving from a source located in a reverberant room at intermediate distances (1-6 m) influences the perception of the distance to the source. First, we conducted an ADP experiment using pure tones (the simplest possible spectrum) of frequencies 0.5, 1, 2, and 4 kHz. Then, we performed a second ADP experiment with stimuli consisting of continuous broadband and bandpass-filtered (with center frequencies of 0.5, 1.5, and 4 kHz and bandwidths of 1/12, 1/3, and 1.5 octave) pink-noise clips. Our results showed an effect of the stimulus frequency on the perceived distance both for pure tones and filtered noise bands: ADP was less accurate for stimuli containing energy only in the low-frequency range. Analysis of the frequency response of the room showed that the low accuracy observed for low-frequency stimuli can be explained by the presence of sparse modal resonances in the low-frequency region of the spectrum, which induced a non-monotonic relationship between binaural intensity and source distance. The results obtained in the second experiment suggest that ADP can also be affected by stimulus bandwidth but in a less straightforward way (i.e., depending on the center frequency, increasing stimulus bandwidth could have different effects). Finally, the analysis of the acoustical cues suggests that listeners judged source distance using mainly changes in the overall intensity of the auditory stimulus with distance rather than the direct-to-reverberant energy ratio, even for low-frequency noise bands (which typically induce high amount of reverberation). The results obtained in this study show that, depending on the spectrum of the auditory stimulus, reverberation can degrade ADP rather than improve it.

Project description:BackgroundExposure to elevated sound pressure levels within the intensive care unit is known to negatively affect patient and staff health. In the past, interventions to address this problem have been unsuccessful as there is no conclusive evidence on the severity of each sound source and their role on the overall sound pressure levels. Therefore, the goal of the study was to perform a continuous 1 week recording to characterize the sound pressure levels and identify negative sound sources in this setting.MethodsIn this prospective, systematic, and quantitative observational study, the sound pressure levels and sound sources were continuously recorded in a mixed medical-surgical intensive care unit over 1 week. Measurements were conducted using four sound level meters and a human observer present in the room noting all sound sources arising from two beds.ResultsThe mean 8 h sound pressure level was significantly higher during the day (52.01 ± 1.75 dBA) and evening (50.92 ± 1.66 dBA) shifts than during the night shift (47.57 ± 2.23; F(2, 19) = 11.80, p < 0.001). No significant difference was found in the maximum and minimum mean 8 h sound pressure levels between the work shifts. However, there was a significant difference between the two beds in the based on location during the day (F(3, 28) = 3.91, p = 0.0189) and evening (F(3, 24) = 5.66, p = 0.00445) shifts. Cleaning of the patient area, admission and discharge activities, and renal interventions (e.g., dialysis) contributed the most to the overall sound pressure levels, with staff talking occurring most frequently.ConclusionOur study was able to identify that continuous maintenance of the patient area, patient admission and discharge, and renal interventions were responsible for the greatest contribution to the sound pressure levels. Moreover, while staff talking was not found to significantly contribute to the sound pressure levels, it was found to be the most frequently occurring activity which may indirectly influence patient wellbeing. Overall, identifying these sound sources can have a meaningful impact on patients and staff by identifying targets for future interventions, thus leading to a healthier environment.

Dataset Information

Segregating complex sound sources through temporal coherence.

Publications

Segregating complex sound sources through temporal coherence.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets