Noise Reduction and Compression Combined?
Brons et al (2015) examined whether dynamic range compression and noise reduction work in tandem or in opposition to each other. They report that the literature on the interaction of these two strategies is scarce. The role of noise reduction is to “reduce hearing aid gain for background noises, while preserving gain for speech…” and the role of dynamic compression is to alter the hearing aid gain, based on the input level so as to fit sound levels into the dynamic range of the wearer. Brons et al report that the literature describing compression is reasonably well known and relatively easy to find, yet the literature on noise reduction is often limited and “is commonly presented as a black box….”
Recordings of four different hearing aid (Starkey, ReSound, Widex, and Phonak) outputs, based on an input of speech in babble at +4 dB signal-to-noise ratio (SBR) were obtained across three experiments—(1)acoustical measurements, (2) perceptual measurements from 16 people with hearing loss to determine “detectability” of processing differences, and (3) perceptual effects from the participants to determine the effect of processing (with regard to speech intelligibility, annoyance of noise, the naturalness of speech and overall preference).
The authors reported processing strategies for noise reduction and compression varied across hearing aids. The strongest effect of combining compression and noise reduction was reduced noise and speech levels. Detectability of the combined processing strategies was different across different hearing aids. The authors state that “…there are relatively large differences between the test hearing aids in terms of gain changes due to noise reduction, compression, and their combined processing….” However, noise reduction and compression did not cancel each other out (in the hearing aids tested), yet speech and noise levels were both reduced. They report compression protocols by themselves were not detectably different, the combined strategies “did not influence speech intelligibility” and they report preference for combined strategies “was lower than previously observed for noise reduction without compression.”
Brons et al state that combined processing (noise-reduction and dynamic-range compression) reduced noise annoyance and did not impact speech intelligibility. They concluded that “conditions with combined processing are not significantly preferred over unprocessed conditions, most probably because compression reduces SNR for input signals with positive SNR. The influence of compression should therefore be considered for the development and evaluation of noise reduction algorithms for hearing aid applications.”
For More Information, References, and Recommendations
Brons I, Houben R, Dreschler AW. (2015) Acoustical and Perceptual Comparison of Noise Reduction and Compression in Hearing Aids. Journal of Speech, Language, and Hearing Research 58(8):1363-1376.