Just Noticeable and Objectionable Group Delays in Digital Hearing Aids

Abstract

Group delay in a digital signal processing (DSP) hearing aid may be perceived as an echo in the sound heard by a wearer listening to his or her own voice, due to a combination of unprocessed sound received at the ear through head and air pathways and delayed sound reaching the eardrum through the hearing aid. Depending on the amount, this delay may be audible or objectionable and can even result in auditory confusion. This study presents results from 18 subjects listening to their own voices through a DSP hearing aid with a variable group delay. The subjects varied the length of the delay and determined the amounts that were noticeable and objectionable as compared to the undelayed condition, while listening to their own amplified voices. Results indicated that a delay of 3 to 5 msec was noticeable to the listeners in 76 percent of the trials, and a delay of longer than 10 msec was objectionable 90 percent of the time. Abbreviations: DAF = delayed auditory feedback, DSP = digital signal processing, KEMAR = Knowles Electronics manikin for acoustic research

Sound Localization Bias and Error in Bimodal Listeners Improve Instantaneously When the Device Delay Mismatch Is Reduced

“…In such cases, the HA instead of the CI stimulation must be delayed to reduce the device delay mismatch. However, this approach has limitations, as HA processing latencies above 10 ms have been shown to cause subjective disturbances in patients (Agnew & Thornton, 2000; Bramsløw, 2010; Groth & Søndergaard, 2004).…”

Measuring and Modeling Cue Dependent Spatial Release from Masking in the Presence of Typical Delays in the Treatment of Hearing Loss

“…Timing differs due to processing delays but also due to physiologic latencies of acoustic hearing introduced by the ear canal, middle ear and mainly the cochlear delays in the inner ear (Ruggero & Temchin, 2007), which are not present when the auditory nerve is directly excited by electric current in the case of CI stimulation. The impact of this temporal alteration on subjective disturbance or sound quality has been studied extensively for provision with HAs (Agnew & Thornton, 2000;Bramsløw, 2010;Groth & Søndergaard, 2004;Stone & Moore, 2003;Stone, Moore, Meisenbacher, 1 Peter Osypka Institute of Medical Engineering, Faculty of Electrical Engineering, Medical Engineering and Computer Sciences, University of Applied Sciences Offenburg 2 Bio-Inspired Information Processing, Munich Institute of Biomedical Engineering, Technical University of Munich & Derleth, 2008). However, most of the studies were looking at maximal preferable delays and did not take the possibility of asymmetric timing between both ears into account.…”

Causal and Relaxed-Distortionless Response Beamforming for Online Target Source Extraction

“…In detail, the delay usually becomes longer than 100 ms because a window longer than 100 ms is widely used in STFT to model reverberation by the instantaneous mixing process in (2). Such a long delay, however, is not acceptable in various real-time applications, including hearing aids [18], [19].…”

“…As users of assistive hearing devices hear both unprocessed and processed signals simultaneously, a delay of the processed signal deteriorates the quality of the heard signal. For instance, listeners without hearing loss can notice a delay of 3 ms and feel uncomfortable with a delay longer than 10 ms [18]. Although listeners with hearing loss tend to tolerate a longer delay, a delay longer than 6 ms is noticeable when they use open-fitting hearing aids [19].…”