Studies have repeatedly documented that many workers exposed to noise are unwilling to wear hearing protection because of fear of failing to understand speech or hear warning sounds. During the last decade, specialized hearing protectors (HPDs) have been developed to address issues related to overprotection from noise and communication. Most employ electronically-modulated sound transmission through the HPD and for this purpose include a microphone outside the device, processing electronics, and a miniature earphone loudspeaker located in or close to the ear (i.e., under the HPD), to transmit sounds from the environment to the user. In many situations requiring the use of HPDs, speech or warning sounds (e.g., back-up alarm, fire alarm), will not be sufficiently loud to support face-to-face conversation or alarm detection. The methods proposed here are intended to improve communication in these circumstances. A recent study has reported a method for reducing noise when attempting to communicate face- to-face in a noisy environment (Lezzoum et al., 2016). It involved first dividing the frequency spectrum of the sounds (i.e., speech + noise) into narrow bands of frequencies (sub-bands). The instantaneous magnitudes of the intensity modulations in each sub-band were then used to adjust the instantaneous gain applied to the signal. In preliminary work for this proposal, we have extended this approach to improve the intelligibility of speech, and the audibility of warning sounds, in noise at the adverse signal-to-noise ratios (SNRs) occurring in occupational environments. We therefore propose to undertake a rigorous study to find and optimize the most suitable modulation-based method for improving speech understanding during face-to-face conversation and alarm perception using sounds recorded in industry, machinery rooms and on construction sites. The speech, alarms and occupational noises will be processed by computer off-line. The sounds will subsequently be presented at different SNRs to subjects in psychoacoustic tests to compare the performance when using processed and unprocessed sounds. Confirmation of a successful method would enable the development of improved electronic HPDs to address the most common complaints of users of hearing protection. With approximately 4.5 million workers in the U.S. unwilling to use HPDs at present, there is reason to expect substantial benefits to public and occupational health from the proposed research for all of NIOSH's Sector Programs in which hearing loss may occur (Agriculture, Forestry, Construction, Manufacturing, Mining, Oil and Gas Extraction, and Transportation). As proposed, the project is focused on NIOSH'S r2P initiative, and falls within the Hearing Loss Prevention, Personal Protective Technology, and Prevention through Design Cross-Sector Programs.
The goals of the research are to improve the intelligibility of speech and the audibility of warning sounds for persons in a noisy environment who are required to wear hearing protection. The method for distinguishing desired sounds buried in noise, as they would be in a noisy occupational environment, from the noise involves electronically processing the sounds. Developing and evaluating the algorithms best suited to these purposes are the subjects of this proposal.
