Acoustic Background Noise Variation in Air Force Platforms and Its Effect on Noise Removal Algorithms
Abstract
In this study of short-term noise variation in Air Force platforms, we followed two avenues of investigation. First, we applied quantitative measures of variation to individual noise recordings, and compared the results across various aircraft. Some measures used were simple descriptive statistics, but we also measured attenuation obtained by spectral restoration (spectral subtraction), applied to the noise signal alone. The noise attenuation obtained for real aircraft environments was in most cases about the same as predicted theoretically for white Gaussian noise, but in some instances was considerably higher, especially in the presence of propeller noise. Second, we applied the nonparametric Mann-Whitney statistic to test the stationarity of power spectrum estimates on time scales of 200 to 800 ms. There was little or no evidence of nonstationarity in large jet or turboprop aircraft. In fighter aircraft and helicopters, there was some evidence of nonstationarity confined to more or less narrow frequency ranges. The nonstationarity found did not appear to limit the performance of special restoration algorithms. The noise recordings used were taken from the RADC/EEV database of field recordings made in the E-3A, E-4B, EC- 135, E-130, P-3C, F-15, F-16, F-4, A-10, HH-53 and Tornado aircraft.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jun 01, 1991
- Accession Number
- ADA238279
Entities
People
- Philip A. Lafollette