^{1,a)}, Jay H. Kim

^{1,b)}, Won Joon Song

^{1}, William J. Murphy

^{2}and Seongho Song

^{3}

### Abstract

Many noise guidelines currently use -weighted equivalent sound pressure level as the noise metric and the equal energy hypothesis to assess the risk of occupational noises. Because of the time-averaging effect involved with the procedure, the current guidelines may significantly underestimate the risk associated with complex noises. This study develops and evaluates several new noise metrics for more accurate assessment of exposure risks to complex and impulsive noises. The analytic wavelet transform was used to obtain time-frequency characteristics of the noise. 6 basic, unique metric forms that reflect the time-frequency characteristics were developed, from which 14 noise metrics were derived. The noise metrics were evaluated utilizing existing animal test data that were obtained by exposing 23 groups of chinchillas to, respectively, different types of noise. Correlations of the metrics with the hearing losses observed in chinchillas were compared and the most promising noise metric was identified.

This project was supported by the National Institute for Occupational Safety and Health, Grant No. R21 OH008510. The authors thank Roger Hamernik and Wei Qiu at the State University of New York at Plattsburgh for providing chinchilla noise exposure study data and advice in interpreting the data.

I. INTRODUCTION

II. METHODS

A. Chinchilla noise exposure test data

B. Analytic wavelet transform characterization of noise

III. DESIGN OF NOISE METRICS

IV. ANALYSIS OF PERMANENT THRESHOLD SHIFTS WITH THE PROPOSED NOISE METRICS

V. CORRELATION STUDY OF METRICS

A. Frequency correlations

B. Noise correlations and overall correlations

VI. DISCUSSION AND CONCLUSION

### Key Topics

- Sound pressure
- 29.0
- Acoustic noise
- 9.0
- Deafness
- 8.0
- Wavelet transform
- 7.0
- Noise propagation
- 6.0

## Figures

Pressure time history of noise G263 for a 3-s period selected randomly out of 5 min period recorded.

Pressure time history of noise G263 for a 3-s period selected randomly out of 5 min period recorded.

T-F representation of the pressure time history shown in Fig. 1 obtained by AWT. The height of the surface at a given frequency-time point indicates the SPL of the 1/3 octave frequency component centered at the frequency and at the time instant.

T-F representation of the pressure time history shown in Fig. 1 obtained by AWT. The height of the surface at a given frequency-time point indicates the SPL of the 1/3 octave frequency component centered at the frequency and at the time instant.

Time histories of the 1/3 octave SPL components of noise G253. Each time history was obtained by applying the AWT to the noise with the center frequency at the frequency shown in the figure. For example, 0.5 kHz time history shown in the figure approximates the G263 noise that passed through a 1/3 octave filter of 0.5 kHz center frequency.

Time histories of the 1/3 octave SPL components of noise G253. Each time history was obtained by applying the AWT to the noise with the center frequency at the frequency shown in the figure. For example, 0.5 kHz time history shown in the figure approximates the G263 noise that passed through a 1/3 octave filter of 0.5 kHz center frequency.

Comparison of the median value of the measured PTS of the animals in group G263 and metric values calculated for the noise at six frequency points. Comparisons are shown for , kurtosis, and (top, from left to right), and (*K*=2), , and (bottom, from left to right). Solid line indicates measured PTS and dashed line indicates metric values.

Comparison of the median value of the measured PTS of the animals in group G263 and metric values calculated for the noise at six frequency points. Comparisons are shown for , kurtosis, and (top, from left to right), and (*K*=2), , and (bottom, from left to right). Solid line indicates measured PTS and dashed line indicates metric values.

Box plots of measured PTS of 18 groups of chinchillas at 0.5, 1, 2, 4, 8, and 16 kHz. Boxes represent inter-quartile ranges, horizontal lines represent the median, whiskers represent the largest and smallest values, and symbols represent outliers defined as the points outside of 1.5 box lengths from the end of the boxes.

Box plots of measured PTS of 18 groups of chinchillas at 0.5, 1, 2, 4, 8, and 16 kHz. Boxes represent inter-quartile ranges, horizontal lines represent the median, whiskers represent the largest and smallest values, and symbols represent outliers defined as the points outside of 1.5 box lengths from the end of the boxes.

Frequency correlation of noise metrics and measured PTS as functions of frequency. The size of a dot represents the correlation values as the scale shows. For example, the dot corresponding to noise index 5 (G252) and metric index 9 represents the frequency correlation of in animal group G252. The two metrics that showed the highest average of the frequency correlations are metric 1 and metric 9 . x indicates a negative correlation.

Frequency correlation of noise metrics and measured PTS as functions of frequency. The size of a dot represents the correlation values as the scale shows. For example, the dot corresponding to noise index 5 (G252) and metric index 9 represents the frequency correlation of in animal group G252. The two metrics that showed the highest average of the frequency correlations are metric 1 and metric 9 . x indicates a negative correlation.

scatter plots showing the overall correlation of : (a) plot of the initial data (108 points:18 groups, 6 frequency points) and (b) plot of the expanded data set (138 points: 23 groups, 6 frequency points).

scatter plots showing the overall correlation of : (a) plot of the initial data (108 points:18 groups, 6 frequency points) and (b) plot of the expanded data set (138 points: 23 groups, 6 frequency points).

## Tables

Description of the noises that were used as the initial set of data for the correlation study in this paper. The data are from the chinchilla noise exposure studies conducted by Hamernik *et al.* (2003a, 2003b) and Qiu *et al.* (2006).

Description of the noises that were used as the initial set of data for the correlation study in this paper. The data are from the chinchilla noise exposure studies conducted by Hamernik *et al.* (2003a, 2003b) and Qiu *et al.* (2006).

Noise correlations of 14 metrics calculated by using the initial exposure data set composed of 18 animal groups exposed to 100-dBA noises. Each correlation is the correlation between the values of the given metric calculated for 18 noises and the PTS values of the 18 animal groups at each frequency. Notice that all metrics have very poor noise correlations at 0.5 and 2 kHz.

Noise correlations of 14 metrics calculated by using the initial exposure data set composed of 18 animal groups exposed to 100-dBA noises. Each correlation is the correlation between the values of the given metric calculated for 18 noises and the PTS values of the 18 animal groups at each frequency. Notice that all metrics have very poor noise correlations at 0.5 and 2 kHz.

Description of the noises that were used as the additional data for the correlation study in this paper. The data are from a new, unpublished chinchilla noise exposure study conducted by Hamernik and Qiu (2001).

Description of the noises that were used as the additional data for the correlation study in this paper. The data are from a new, unpublished chinchilla noise exposure study conducted by Hamernik and Qiu (2001).

Noise correlations of 14 metrics calculated by using the expanded exposure data. The expanded data were obtained by adding the data from three animal groups exposed to 90-dBA noises and two animal groups exposed to 90-dBA noises. Correlations have become higher in general, especially at 0.5 and 4 kHz.

Noise correlations of 14 metrics calculated by using the expanded exposure data. The expanded data were obtained by adding the data from three animal groups exposed to 90-dBA noises and two animal groups exposed to 90-dBA noises. Correlations have become higher in general, especially at 0.5 and 4 kHz.

Article metrics loading...

Full text loading...

Commenting has been disabled for this content