Skip directly to search Skip directly to A to Z list Skip directly to navigation Skip directly to page options Skip directly to site content


workers, building, architect

NORA Manufacturing Sector Strategic Goals

927ZJNG - Testing Speech Intelligibility and Hearing Protection

Start Date: 4/1/2010
End Date: 9/30/2013

Principal Investigator (PI)
Name: David Byrne
Phone: 412-386-6576
Organization: NIOSH
Sub-Unit: DART
Funded By: NIOSH

Primary Goal Addressed

Secondary Goal Addressed


Attributed to Manufacturing


Project Description

Short Summary

The objective of this research is to investigate the effects that hearing protectors have on verbal communication in the workplace, and specifically to evaluate the best method to assess these effects. Development of a new speech intelligibility measure is necessary, since all existing test materials are based on speech produced in quiet conditions, which is not representative of a workplace with high ambient noise levels. The Construction and Manufacturing sectors and Hearing Loss Prevention cross sectors will benefit from this research. This project will have application to the millions of workers who are exposed to noise in their occupations. The ability to identify appropriate protection that maximizes the ability to communicate in a noisy environment is a critical need listed in the NIOSH Hearing Loss Research Strategic Plan. Existing speech intelligibility tests will be integrated to develop new a test method and materials specifically tailored for use with hearing protection devices. Consequently, the translation of this new method will be developed into a new standard and will become a part of an updated best practices document for preventing occupational hearing loss.


During the first year of the project, attention will be focused on assessing the effect of earcanal occlusion, hearing protector attenuation, and background noise level on the intensity and frequency content of a talker�s vocal output. To test whether differences between occluded and unoccluded vocal characteristics are due solely to the occlusion effect, speech produced while subjects� earcanals are occluded will be measured without the subject receiving any attenuation from the hearing protectors. To achieve this condition, the subject�s own voice will be reproduced (in real-time) through a set of headphones, essentially restoring the same sound level reaching his/her ears as when the hearing protectors are not worn (i.e., the attenuation of the protectors is being offset). To test whether vocal output differences are due to the reduction in the talker�s self-perceived voice level, the amount of occlusion will be held constant while changing the effective hearing protector attenuation. Findings from this preliminary work will be used as the basis for determining the appropriate recording parameters for subsequent parts of the project.

During the second year of the project, test materials (i.e., word/sentence lists) will be recorded when spoken by a professional voice talent. A contract with a local agency/individual will be established, and the recording will be conducted in the laboratory facilities at the NIOSH Robert A. Taft building. Three different test formats will be used: single words; sentences; and connected discourse. All 300 words from the Modified Rhyme Test will be included. Sentences will be those obtained from the IEEE speech quality measurement work, and as found in the pre-recorded QuickSIN� test. A 25-word passage from the Connected Speech Test will be used for the connected discourse testing. The talker (i.e., professional announcer) will be fitted with each set of earplugs, and will be positioned in a double-wall sound treated audiometric booth. He will be instructed to read the test materials in quiet and increasing pink noise levels (75-95 dBA) in 10-dB increments. The talker�s voice will be recorded with the measurement microphone located 1 meter directly in front of his mouth, while the background noise will be delivered via calibrated headphones. The talker will be monitored to ensure the expected vocal characteristics are produced, i.e., those based on the results from the earlier part of this study.

The third year of this project will be used to conduct listening tests where the (pre-recorded) talker as well as the listener is subjected to background noise and both are wearing hearing protection. While seated in the sound-treated audiometric booth and fitted with earplugs, subjects will be instructed to listen and repeat back (without the benefit of visual cues or other assistance) each word/sentence that they hear. The word/sentence lists will be presented at a level equivalent to that at which they were recorded (as described above), while the background noise is played through a single loudspeaker located directly in front of the subject. Words will be scored individually, and for each sentence list the number of words correctly repeated divided by the total words in each list (and multiplied by 100) will indicate the percent correct. The connected discourse testing will be scored by having the subjects answer five questions directly related to the subject matter presented to them. All subject responses will be digitally recorded to facilitate double-checking the results at a later time, if necessary.

At the conclusion of this study, the type of test that produces the widest distribution (i.e., greatest separation) of intelligibility scores across the different listening conditions will be identified. Results obtained from this project will provide more insight into the type of test best-suited for assessing speech intelligibility in noise. Subsequent analysis will enable the development of recommendations regarding the use of a particular speech intelligibility test in actual workplace Hearing Loss Prevention Programs.

Mission Relevance

An estimated four million U.S. workers are exposed to hazardous noise on a daily basis. This presents a significant problem that extends across nearly every occupational sector. Noise-induced hearing loss is virtually always preventable if suitable hearing protection is used. However, workers refuse to wear hearing protection because they complain it interferes with normal speech communication.

Existing research has shown that wearing hearing protection does not prevent normal-hearing individuals from understanding speech in noise levels above 85 dB(A). In fact, due to inherent physiological phenomena, normal-hearing workers may actually hear speech better when wearing earmuffs or earplugs in a noisy work environment. Unfortunately, data regarding the effects of a particular hearing protector on speech intelligibility are not available from the manufacturer. Therefore, on-site safety or hearing conservation program managers typically do not know to what extent each device affects speech intelligibility for a given acoustic environment and employee�s hearing status. Consequently, potential communication issues cannot be identified when hearing protectors are selected. Current hearing conservation standards, regulations, and practices do not address this problem.

Commonly used clinical speech-in-noise tests do not necessarily provide insight into identifying which hearing protectors might be better for a particular individual in a given noise/communication situation due to two important limitations: First, the acoustical characteristics of speech produced in a noisy work environment will be different than the pre-recorded speech material found on commercially available tests. Second, non-acoustic factors such as linguistic structure and message content/predictability are quite different between pre-recorded tests and actual workplace communications. The intent of this project will be to research these basic underlying factors to identify/develop a single test that will be accepted as the �standard� for evaluating the effect a hearing protector has on communication in noise.