Authors: Lucas Mens1 , Arno Janssen1 , Jan-Willem Wasmann1
1Department of Otorhinolaryngology, Donders Institute for Brain, Cognition and Behaviour,
Radboud University Medical Centre Nijmegen, the Netherlands
Background: The rationale for developing the at simple at-home self-check is twofold. 1) Fitting procedures for hearing prostheses focus on pure-tone thresholds (hearing aids) or most comfortable loudness levels measured per electrode (cochlear implants) ignoring several factors affecting the loudness of complex signals such as speech, the spatial and temporal integration, and the signal pre-processing. Ignoring above factors can lead to a suboptimal loudness balance and/or uncomfortable loud sensations at high input levels. We present a simple listening test using complex signals that has proven to be effective in fitting CI users at our center in the past decade. 2) Cochlear implant centers are confronted with an increasing number of patients in life-long care increasing the need to reduce or eliminate check-up visits. A concept of a self-test version of the clinical listening test will be presented, potentially helpful in reassuring patients that all is well or detecting an actual need for a visit or remote counseling.
Method: The listening test uses the filtered “Ling” sounds /oo/, /aa/, /ee/, /sj/ and /s/. It’s a set of stimuli that covers the whole speech spectrum. Each stimulus is presented at a conversational and a raised level. Sounds are filtered to be mutually non-overlapping and each sound therefore stimulates a known set of electrodes. A loudness scale is used to collect responses. LDL responses to the raised level sounds can be addressed by reducing the MCL of the corresponding set of electrodes. Target loudness levels are defined for the sounds at conversational levels and corrections are made by changing upper and lower stimulation levels, and possibly also input gain settings. In addition, the pitch variation between the sounds can be assessed as a crude evaluation of tonotopy. A preliminary version of a self-test tool is designed. Sound input via streaming and via the microphone can be compared to coarsely detect microphone degradation.
Figure showing one of 5 filtered speech sounds (/AA/), it’s bandwidth (580 – 1650Hz) and the default set of electrodes of a particular CI brand that correspond with that bandwidth.
Results: The listening test is part of our clinical routine for Ci fittings since many years. However, no formal evaluation has been attempted. A case will be presented in which this test was critical to arrive at a fitting. The self-test version is being used in a preliminary field test.
Conclusion: A measure of the loudness of speech-like sounds has been very useful in our CI clinic. Once available as an app for mobile devices, it can easily been used – in addition to tests of speech understanding – to have patients evaluate their fitting (cochlear implant or conventional hearing aid) and is expected to support remote counseling and the decision if a visit in the clinic is required or can be postponed.
I wonder if whole word or sentence loudness is well predicted by these filtered speech sounds?
Hi Dave,
You are absolutely right. What I did not mention in the abstract is that I have repeated each phoneme 9 times with very short intervals, mimicking the temporal pattern of a sentence (by a stutterer, as it were). Of course, being filtered, these “sentences” contain less energy than normal speech, but thus seem to serve the purpose well of approaching the temporal pattern of running speech while maintaining a known relation with a subset of channels.
(The overall sentence loudness remains under the control of the user as far as the volume setting allows.)