We included 11 individuals with a dry single sided radical cavity. Of these 11 individuals, 6 (55%) were female. The average age of all participants was 43.6 years (median 39, range 25–64 years). Seven individuals (63%) have a right sided cavity. The subjects had a cavity ear for an average of 21 years (median 26, range 7–33 years). Of these cavity ears, 2 (18%) had a history of revision surgery of their cavity (Table 1). Bone conductive tone thresholds of the cavity ear were 20 dB HL or better at 0.25, 0.5, 1, 2 and 4 dB and air conductive tone thresholds were better than 70 dB HL at 0.25, 0.5, 1, 2 and 4 dB. Pure air conductive thresholds of the contralateral ear were 20 dB HL or better at 0.25, 0.5, 1, 2 and 4 dB. All participants agreed to participate in the study. The study protocol was in accordance of the Helsinki declaration and was approved by the ethical review board. None of the authors has a conflict of interest.

In short, we created filtered sound fragments for each participant that simulate the different acoustic properties of the cavity ear and the contralateral normal ear canal based on the participants individual Real Ear to Coupler Difference (RECD). These filtered sound fragments are presented to the cavity ear canal and the contralateral ear canal. Participants performed a subjective quality assessment of the filtered sound fragments using paired comparison.

To acquire the individual filtered sound fragments, we measured acoustic properties of the individuals ear canals via the RECD, being the frequency dependent gain in decibels (dB) of the soundwave from the insert microphone to the eardrum [12]. The RECD is measured using a microphone inserted in the ear canal, placed near the eardrum, measuring the frequency dependent gain of the soundwave in decibels of a well-defined broadband sound stimulus from the foam insert microphone in the ear canal (Fig. 1). All ear canals were dry and properly cleaned before measurement.

In this figure, a schematic representation of the real ear to coupler (RECD) measurement is displayed. The RECD accounts for the decibel difference across frequencies between sound pressure level (SPL) measured in the 2 cc coupler and the SPL measure in the participants individual ear, produced by the same transducer generating the same soundwave in decibels of a well-defined broadband sound stimulus. SPLM sound pressure level measure

Differences between individual RECDs represent differences in acoustic properties of individual ear canals. The acoustic effect of the measured acoustic properties of different ear canals can be simulated in the participants using a filter on the incoming sound stimulus. Therefore, we use the difference between the RECD of the individuals normal ear canal and the RECD of the cavity ear canal, using the REM module of the Affinity 2.0 Hearing Aid Analyzer platform (Interacoustics, Denmark). For our participants, this filtering results in the same distribution of sound pressure (acoustics) at the eardrum as in the contralateral ear canal, mimicking the acoustic effect of the cavity in a normal ear, and mimicking the acoustic effect of the normal ear in the cavity ear [6, 7, 10].

We made recordings of Dutch speech (two male and two female speaker sentences based on the VU98 sentence material [13]), filtered to simulate the acoustic properties of the individuals contralateral ear canal. The acoustic filters, being the simulated conditions, were built on the differences between the individuals RECDs (Fig. 2).

These graphs illustrate how we created our sound filters. In the top chart, the RECD of the participants normal ear canal is displayed in green and the participants cavity ear canal in red. In the bottom chart the green line displays the difference in RECD between the cavity canal and the normal ear canal. In red, the difference in RECD is smoothed. The blue line is the frequency dependent filter we used to mimic the cavity canal acoustics in the normal ear of this participant. The inverse of this filter is used to mimic the frequency dependent filter to mimic normal ear canal acoustics in the cavity ear canal. These filters are created for each participant. RECD real ear to coupler difference

The perceptual evaluation experiment was performed with a paired comparison category rating between two fragments (‘A’ and ‘B’), according to ITU-T 1996 [14]. Participants were asked for each ear which fragment sounded more natural, more clear and better using a seven-point scale. Each filtered cavity sound fragment was compared to the sound fragment of the normal ear. These sound fragments are presented and compared, first in the normal ear and then in the cavity ear, using a foam insert microphone. All conditions were presented using four (Dutch) sentences by two male and two female voices and were measured twice for each: one time using the cavity filtered sound fragment as ‘A’ and the normal ear sound fragment as ‘B’, and one time in a reversed fashion. Thus, eight paired comparisons were presented in random order to assess naturalness of the sound for each ear canal. The same 8 paired comparisons were presented in random order to asses clearness of the sound, 8 paired comparisons to asses overall quality of the sound fragment, making a total of 48 paired comparisons for both ears. The paired comparison measurements are stated on a seven-point scale ranging from + 3 (the filtered cavity sentence sounds much more natural, clear or better than the normal ear) to − 3 (the normal ear sentence sounds much more natural, clear or better than the filtered cavity sentence). A score of 0 means there is no difference noticeable in naturalness, clearness or quality of sound.

All of the speech material was presented using in ear inserts at a level of 65 dB(A). In the cavity ear canal the microphone is amplified according to the NAL-NL1 rule to simulate the perception of 65 dB. Sound fragments were assigned to the subject in a random order alternately presented to the cavity ear canal and the normal ear canal.

Data are expressed as numbers. Naturalness, clearness and quality of sound fragments presented to the normal ear and the cavity ear were compared using paired two tailed sample t test using IBM SPSS Statistics 26.