Optimizing Soft Speech Recognition in Children With Hearing Loss (SoftSpeech)

Overhearing is important for vocabulary learning and speech and language development in young children. However, contemporary hearing aids are generally unable provide adequate access to low-level auditory inputs from multiple talkers at a distance to capitalize on overhearing. A recent investigation by Jace Wolfe and colleagues showed that, even when aided, children with hearing loss had significantly poorer speech recognition at 40, 50 and 60 dBA compared to children with normal hearing. Furthermore, they showed that increasing hearing aid gain for very low-level inputs produced a statistically significant improvement in syllable-final plural recognition and a non-significant trend toward better monosyllabic word recognition at very low presentation levels. Additional research is needed to document low-level speech recognition ability of children with hearing loss as well as the potential benefit or detriment of increasing hearing aid gain for low-level inputs. A novel hearing aid technology known as Soft Speech Enhancer has been shown improve low-level speech perception in adults with hearing loss; however, the effect of Speech Enhancer on speech recognition in children is not yet known and will be evaluated.

The primary objective of this study is to evaluate whether soft speech recognition in quiet is significantly improved with Speech Enhancer (SE) "on" at the default setting (moderate) compared to Speech Enhancer "off" in children with moderate (N3) to severe (N5) hearing loss. There are several secondary objectives: 1) To evaluate the main effects of age and Speech Enhancer strength, and potential interaction on speech recognition and comfort in children with moderate to severe hearing loss. 2) To confirm no negative effect of speech enhancer "on" at the default setting on low-level speech recognition in noise. 3) To demonstrate that hearing aids compensate for hearing loss in children. This confirmatory clinical investigation will be executed at Hearts for Hearing. It is a single-group, single blind intervention study with each participant serving as his or her own control. However, in certain conditions participant blinding is not feasible (i.e., aided versus unaided performance). The outcome measures being assessed include: speech recognition in quiet and noise with CNC words, speech detection in quiet with the UWO Plurals test, comfort ratings using a Multiple Stimuli with Hidden Reference and Anchor (MUSHRA) procedure and looped sentences overlaid with low-level transient noise. Four different hearing aid conditions will be assessed in this clinical investigation: unaided (no amplification), aided with SE "off", aided with SE "moderate", aided with SE "strong". For the unaided or no treatment condition no hearing aids will be worn. For aided conditions, experienced hearing aid users will be fit binaurally with investigational devices matching their personal device form factor connected to their personal acoustic coupling (e.g., earmold). Using the Audioscan Verifit 2, probe microphone measurements will be conducted to match DSL 5.0 pediatric quiet and noise targets. Three distinct programs will be created, each with a different SE strength. Hearing aid program order (speech enhancer settings) will be randomized for all testing conditions except for CNC at 50 dBA Quiet with SE Off; that condition is administered first to establish candidacy for further testing. A randomized list of SE settings was generated for all participants for the remaining conditions. Testing will be completed in a double-walled sound booth with calibrated equipment. Recorded stimuli will be presented through loudspeakers (Sony CFD-ZW755) from clinical audiometers (Grayson Stadler Standard) or Dell computers using the Arizona State University Speech Recognition Program, Windows Media Player and Adobe Audition. Speech stimuli will be presented from a loudspeaker located at 0 degrees azimuth (directly in front) and noise will be generated from a single loudspeaker located at 180 degrees azimuth (directly behind) the patient. Speech recognition results will be compared to determine the effect of amplification on accuracy scores as well as the influence of speech enhancer on behaviorally measured benefit and subjectively-derived listening preference.

One group of participants will be assessed unaided and aided with Phonak Audeo P or Phonak Naida P hearing aids with Speech Enhancer "off", "on at moderate", and "on at strong" settings.

Behavioral and subjective preference data will be collected in children with hearing loss who are unaided (no technology) or aided with Phonak Audeo P or Naida P hearing aids with Speech Enhancer on at moderate, on at strong, or off.

Difference in aided speech recognition accuracy, percent correct CNC word scores, in 44 participants listening with Speech Enhancer "on" at the moderate setting compared to Speech Enhancer "off".

Difference in aided speech recognition accuracy scores, percent correct CNC word scores, between 22 younger (4-8 year old) and 22 older (9-12 year old) children listening with Speech Enhancer activated at "moderate" and "strong settings.

Difference in soft speech recognition scores, measured as percent correct CNC word scores, in 44 patients with hearing loss listening unaided compared to aided with Phonak Audeo P or Naida P devices.

Inclusion Criteria: native English speakers who communicate verbally no known speech language disorders, cognitive or learning issues experienced hearing aid users (> 6 months use) "full-time" hearing aid users (i.e., wear on a daily basis) CNC word scores of 30% or better at 50dBA in quiet with SE "off" Exclusion Criteria: new hearing aid users inconsistent hearing aid users <4 years or >12 years of age non-native English speaker who communicates verbally known cognitive/ learning issues or speech language disorder CNC word scores < 30% correct presented at 50dBA in quiet with SE "off"