Hearing Through Bone Conduction Headsets

Abstract

Bone conduction (BC) stimulation has mainly been used for clinical hearing assessment and hearing aids where stimulation is applied at the mastoid behind the ear. Recently, BC has become popular for communication headsets where the stimulation position often is close to the anterior part of the ear canal opening. The BC sound transmission for this stimulation position is here investigated in 21 participants by ear canal sound pressure measurements and hearing threshold assessment as well as simulations in the LiUHead. The results indicated that a stimulation position close to the ear canal opening improves the sensitivity for BC sound by around 20 dB but by up to 40 dB at some frequencies. The transcranial transmission ranges typically between −40 and −25 dB. This decreased transcranial transmission facilitates saliency of binaural cues and implies that BC headsets are suitable for virtual and augmented reality applications. The findings suggest that with BC stimulation close to the ear canal opening, the sound pressure in the ear canal dominates the perception of BC sound. With this stimulation, the ear canal pathway was estimated to be around 25 dB greater than other contributors, like skull bone vibrations, for hearing BC sound in a healthy ear. This increased contribution from the ear canal sound pressure to BC hearing means that a position close to the ear canal is not appropriate for clinical use since, in such case, a conductive hearing loss affects BC and air conduction thresholds by a similar amount.

Introduction

In recent years, communication headsets that claim to use bone conduction (BC) transmission have become popular (Brown et al., 2019; Lim & Claydon, 2020; Manning et al., 2017). Common for most of these devices is a headband with two BC transducers placed onto the skin in front of the ear canal opening. Such placement is different from classical and well-researched positions such as the position behind the ear canal opening (often referred to as the B71 position) (Dobrev et al., 2016), the forehead position (Khanna et al., 1976), or the position for percutaneous bone conduction implants (often referred to as the BAHA position) (Eeg-Olofsson et al., 2008; Tjellström & Håkansson, 1995). The transducer position for a BC headset is often superior to the area between the condyle of the lower jaw and the tragus. The tissues beneath the BC headset transducer are skin, subcutaneous tissue, cartilage, and bone. A vibration application close to the ear canal and on the cartilage can result in sound transmission to the inner ear that is different from the normally anticipated vibration transmission pathways (Nishimura et al., 2015).

With BC stimulation on the skin-covered skull bone, the sound excites the inner ear via different mechanisms, and these are often described as five different pathways (Stenfelt, 2011, 2020; Stenfelt & Goode, 2005a):

Sound pressure induced in the ear canal that is transmitted to the inner ear via the middle ear ossicles (Stenfelt et al., 2003; Surendran & Stenfelt, 2022).

Inertial effects on the middle ear ossicles causing a relative motion between the stapes footplate and the surrounding bone (Röösli et al., 2012; Stenfelt, 2006; Stenfelt et al., 2002).

Sound pressure created in the cranial cavity that is subsequently transmitted to the inner ear via compliant pathways (Dobrev et al., 2019, 2022; Sohmer et al., 2000; Stenfelt & Prodanovic, 2022).

Inertial effects of the inner ear fluid causing a fluid motion (Kim et al., 2011; Stenfelt, 2015).

Compression and expansion of the inner ear space causing a net fluid flow over the basilar membrane (Stenfelt, 2015, 2020).

Of these, based on model simulations, the inertial effects of the inner ear fluid have been suggested as the most important in a normal healthy ear (Stenfelt, 2016, 2020).

Vibrations that are applied to the auricle are also transmitted as sound to the inner ear (Nishimura et al., 2015; Shimokura et al., 2014). This application of vibration is different from the classical positions where the skull bone is directly under the excitation area, with or without skin and soft tissues in-between. As a result, this transmission has been termed cartilage conduction and has been suggested to be different from conventional BC transmission and air conduction (AC) transmission (Shimokura et al., 2014). However, it is not entirely clear if this cartilage conduction uniquely transmits sound to the inner ear, or is merely BC sound transmission where the ear canal component is enhanced due to radiation from the transducer itself or due to vibration of the ear canal wall (Nishimura et al., 2015).

Four of the five pathways for BC sound transmission described above depend on the skull bone vibration. The inertial effects of the inner ear and the alteration of the inner ear space correlate to the vibration of the cochlear promontory (Stenfelt, 2015), the contribution of middle ear inertia depends on the vibration of the skull bone encapsulating the middle ear (Stenfelt et al., 2002), and the sound pressure in the cerebrospinal fluid is created by the local skull bone vibration (Dobrev et al., 2022; Stenfelt & Prodanovic, 2022). The ear canal sound pressure, as discussed in the previous paragraph, can be partly independent of the skull bone vibration. Consequently, with BC transducer application as typical with BC headsets, the ear canal sound pressure can be enhanced in relation to the skull bone vibration normally associated with BC hearing.

The aim of the study is to investigate the sound transmission from a vibration transducer positioned in front of the ear canal like in most BC headsets. The hypothesis is that with the positioning of the BC transducer in front of the ear canal opening, the ear canal sound pressure is enhanced in relation to other pathways that rely on skull bone vibrations. The sound transmission will be evaluated experimentally and in the computational model LiUHead. In the experiment, the ear canal sound pressure at threshold is evaluated for ipsilateral and contralateral BC stimulation when the BC transducer is in front of the ear canal opening, and when the BC transducer is placed at the mastoid usually used for BC audiometry. The difference in sound pressure at threshold between ipsilateral and contralateral stimulation for the two stimulation positions is used to estimate the importance of the outer ear pathway for BC hearing. In the LiUHead model, the vibration of the cochlear promontory is evaluated for stimulation applied to the skin-covered mastoid and in front of the ear canal opening. The results are compared with data from the experiment to evaluate the ear canal sound pressure at the same cochlear promontory vibration for the two stimulation positions.

Method

Ethics

The current study was approved by the Swedish Ethical Review Authority (2020-01235).

Simulations

The simulations are conducted in the finite element model LiUHead that was developed for BC investigations (Chang et al., 2016). The LiUHead has previously been used to evaluate BC sound transmission as measured at the cochlear promontory (Chang et al., 2016; Prodanovic & Stenfelt, 2021), BC sound power transmission in the head (Chang et al., 2018), BC hearing aids (Chang & Stenfelt, 2019), effects of mastoidectomy on BC sound (Prodanovic & Stenfelt, 2020), and BC transmission with stimulation applied at soft tissue positions (Stenfelt & Prodanovic, 2022). Details of the original LiUHead can be found in Chang et al. (2016) and additional modifications are presented here.

Figure 1A shows a side view of the LiUHead with the two stimulation interfaces and Figure 1B shows a horizontal cross-section of the LiUHead right ear. The stimulation was either applied as a dynamic force over a circular plate of 175 mm² on the mastoid (termed mastoid position) or as a dynamic force over a rectangular-shaped area (195 mm²) with curved corners in front of the ear canal opening (termed frontal position). The mastoid position simulates the application of an audiometric BC transducer, the Radioear B71 transducer (Radioear, DK), similar to the simulations presented in Chang and Stenfelt (2019) and Prodanovic and Stenfelt (2020). It also simulates the BC transducer at the mastoid position in the experimental part of the current study. The mechanical parameters of the soft tissue between the circular plate and the skull bone (orange in Figure 1B) were altered to account for a 5.4 N static force according to the method presented in Chang and Stenfelt (2019) and Prodanovic and Stenfelt (2021). The frontal position simulates the application of a typical BC headset and the interface used here is the same as in the Aftershokz Aeropex (us.aftershokz.com). No alteration of the soft tissue parameters was done to account for the static force at the frontal position.

Figure 1.

(A) Side view of the LiUHead with the two plates for BC stimulation, at the mastoid (circular interface) and at the frontal position (rectangular interface). The red line indicates the plane for the cross-sectional image in (B). (B) Horizontal cross-sectional image of the right ear of the LiUHead showing the two stimulation plates, the ear canal (light blue), soft tissue (pink), cartilage (magenta), cortical bone (gray), CSF (blue), and brain tissue (brown). The orange area under the mastoid stimulation plate indicates the soft tissue with mechanical parameters adjusted for a static force of 5.4 N. The cochlear promontory velocity (V_CP) is indicated by the circle. BC: bone conduction; CSF: cerebrospinal fluid.

The simulations are conducted by applying a uniformly distributed dynamic force of 1 Newton over the stimulation area. The response was extracted as the velocity at the stimulation position and at the cochlear promontory at a bony position between the oval and round window (indicated by V_CP in Figure 1B). The frequency range for the simulations is 100 Hz to 10 kHz.

Participants

Measurements were performed on 21 adult volunteers in the age range 18 to 51 years (mean age 33 years) of whom 7 were female. They all reported subjective normal hearing and no tinnitus problem. They were part of a previous study and the results of the AC hearing and the BC hearing with stimulation on the mastoid are detailed in Surendran and Stenfelt (2022). The ear canal sound pressures with AC and BC mastoid stimulation in Surendran and Stenfelt (2022) are used for comparison to the ear canal sound pressure with stimulation in the frontal position in the current study. The volunteers received 500 Swedish Krona for their participation.

Experimental Setup

All testing was computerized, and the BC sound was provided through a 24-bit 96 kHz external sound card (TASCAM US-16 × 08, TEAC Corp., JP) and amplified by a power amplifier (Rotel RA-04 SE, Rotel Co, Ttd, JP). For BC testing on the mastoid position, a bilateral BC headset was designed based on two motor units from the Radioear B81 BC transducer connected to a Cochlear™ Baha® SoundArc (Cochlear BAS, SE) (Surendran & Stenfelt, 2022). This resulted in a stimulation interface area of approximately 175 mm² at a position close to the ordinary place for a Radioear B71 or B81 transducer with a static force of 2 to 3 N. For BC testing on the frontal position, an Aftershokz Aeropex headset (us.aftershokz.com) was used where the electrical signals from the power amplifier were connected directly to the BC transducers of the headset. The headset also provided the static force that was around 2 N. AC stimulation was provided by a pair of modified ER3 earphones (Etymotic Research Inc, IL) where the foam inserts were reduced to three wings of 2 mm thickness resembling a radioactive symbol. This manipulation of the foam inserts was done to avoid an occlusion effect (Stenfelt & Reinfeldt, 2007).

Ear canal sound pressure was obtained by placing probe tube microphones (ER7C, Etymotic Research Inc, IL) with the openings within 5 mm from the eardrum. The ER7Cs were connected to two inputs of an I/O card (NI USB-4431, National Instruments Corp, TX) that also monitored the electrical signals to the BC transducers, either at the mastoid or frontal position. In this way, the ear canal sound pressures for a given electrical stimulation to the BC transducers were obtained in the frequency range of 0.1 to 12.5 kHz.

Procedure

All testing was conducted in a sound insulated booth where the participant was seated in a chair. Due to practical reasons, the same test order was used for all participants and the left ear was used as the test ear. After placing the probe tube microphones, the AC hearing thresholds with ER3 stimulation were obtained as a function of the ear canal sound pressure measured by the probe tube microphones. Then the ER3 in the left ear was removed while the ER3 with the modified insert in the right ear remained for masking. A foam earplug (3M™ E-A-R™ Classic™, 3 M, MN) was inserted approximately 15 mm into the left ear and the BC transducers were placed on the mastoids. Then the ipsilateral and contralateral occluded ear canal sound pressure with BC stimulation at the mastoid position was obtained from the left ear. Thereafter, the BC transducers at the mastoid position were removed, the Aftershokz transducers were placed in the frontal position, and the occluded ear canal sound pressures with ipsilateral and contralateral BC stimulation were measured. After that, the foam earplug was removed and the open ear canal sound pressure with ipsilateral and contralateral BC stimulation was obtained, first with the BC transducer in the frontal position and then with the BC transducer at the mastoid position.

Then, the hearing thresholds with mastoid stimulation were measured. To ensure that the hearing response was solely from the left ear, a narrow-band noise delivered by the ER3 masked the right ear. The masking noise was centered around the test frequency at a level of 30 dB above the test tone. Finally, the hearing thresholds were measured with BC stimulation in the frontal position using the Aftershokz headset and with contralateral masking.

Hearing thresholds were obtained at one-third octave frequencies between 250 Hz and 12.5 kHz using an adaptive three-alternative forced-choice method. In short, a tone was presented for 1s at one of the three periods. Each period was illuminated on a screen in front of the participant and the time between periods was 0.5 s. If the participant correctly identified the period of the stimulation for two consecutive trials, the stimulation level was reduced, while one erroneous identification increased the stimulation level (two-down one-up algorithm). The testing started at a level of around 40 dB above the expected hearing threshold, and the step size was 10 dB. After two runs (one minima and one maxima) the step size was reduced to 5 dB and after additional two runs the step size was reduced to 1 dB and four runs were completed. The hearing threshold was computed as the mean of the last four turnpoints. Hearing thresholds were obtained with both ipsilateral and contralateral stimulation.

Calibration and Impedance Measurement

The output of the BC transducers was measured on an artificial mastoid Brüel and Kjær type 4930. The mechanical impedance of the artificial mastoid should replicate the mechanical impedance of the human skin-covered mastoid when a circular interface of 175 mm² is applied with a static force of 5.4 N (IEC:60318-6, 2007). However, the impedance level of the artificial mastoid is 2 to 5 dB higher compared to the impedance of the human mastoid. This can cause deviations between the applied output and the estimated output from a BC transducer at the mastoid (Surendran & Stenfelt, 2022). It is expected that the mechanical impedance at the frontal position deviates more from the artificial mastoid than the human mastoid does. This implies that the estimated output using the artificial mastoid would be less accurate for the frontal position compared to the mastoid position. Therefore, the mechanical impedance at the frontal position was measured to assess its impact on the output of the BC transducers.

The assessment of the mechanical impedance was conducted in two ways. The first was to extract the mechanical impedances from the simulations in the LiUHead. The mechanical impedances at the mastoid position (Z_M) and the frontal position (Z_A) are defined as

$$\{\begin{array}{c}{Z}_M={\displaystyle \frac{F_M}{v_M}}\\ Z_A={\displaystyle \frac{F_A}{v_A}}\end{array}$$

(1)

where F_M and v_M are the applied force and response velocity at the mastoid and F_A and v_A are the applied force and response velocity at the frontal position (Figure 1). The response velocities are computed as the mean velocity over the stimulation interface in line with the applied force vector when a dynamic force of 1 N is applied.

The mechanical point impedances were also obtained in six of the participants. This was accomplished with the Brüel and Kjær type 8000 impedance head attached to a Brüel and Kjær type 4810 mini shaker. The mini shaker was mounted on a lever that controlled the static force by a mass. In this way, the mechanical point impedances with a static force of 2 N were measured for a 175 mm² circular area at the mastoid and frontal positions.

To facilitate estimation of the different mechanical impedances’ impact on the BC transducers’ output, the mechanical impedances of the two stimulation positions were mimicked by affixing a silicone pad on the artificial mastoid. Silicone pads of different thicknesses were manufactured and their impedances were measured by the Brüel and Kjær type 8000 impedance head when placed on the Brüel and Kjær type 4930 artificial mastoid. The silicone pads that matched the average impedances from the six participants were used to assess the output from the BC transducers. The output sensitivity of the BC transducer with the silicone pad attached was estimated by obtaining the force transfer function from the pad interface to the force gauge of the artificial mastoid according to Scott et al. (2015).

Results

Mechanical Point Impedance

Figure 2A shows that the simulated mastoid impedance with 5 N static force (blue line) and the median of the measured mastoid impedance with 2 N static force (black line) are within 5 dB at frequencies above 200 Hz. The impedances for the frontal position in Figure 2A are mostly similar, but the simulated impedance is 5 to 15 dB below the measured impedance at frequencies between 200 and 800 Hz. This could partly originate in the different static forces where the simulation on the LiUHead is done with no static force and the experimental data are obtained with 2 N static force. Also, the simulations are done with a 195 mm² rectangular area while the experimental data are obtained with a 175 mm² circular area. The spread of the measured impedance magnitudes for the six participants at 200 to 800 Hz was 3 dB. All impedance magnitudes from simulations and measurements merge at frequencies above 2 kHz indicating an impedance of a pure mass of approximately 1.1 g (indicated by the black dashed line). The low-frequency compliance (inverse of stiffness) is around 8 µm/N for the mastoid (blue dashed line) and 40 µm/N at the frontal position. In addition, the estimated impedance of the artificial mastoid from Surendran and Stenfelt (2022) is included (cyan line). This shows that the artificial mastoid impedance is 15 to 20 dB higher than the impedance at the frontal position for frequencies below 2 kHz and 5 to 10 dB higher than the mastoid impedance for the same frequency range. Consequently, output force assessments of the BC transducers using the artificial mastoid are biased for frequencies below 2 kHz, probably more so for the Aftershokz transducer than for the BC transducer used at the mastoid.

Figure 2.

(A) The mechanical point impedance at the mastoid and the frontal position estimated from simulations in the LiUHead (blue and red lines) as well as measured on humans with 2 N static force (mastoid: black line, frontal position: magenta line). The vertical bars indicate ± 1 SD around the median. The cyan line is an estimation of the mechanical impedance of the artificial mastoid. The impedance of a pure mass of 1.1 g (black dashed line) and two stiffnesses (blue and red dashed lines) are included for comparison. (B) The output force level (dB re 1 µN) measured on the artificial mastoid Brüel & Kjær type 4930 with 1 volt to the transducer. The blue line shows the mastoid-placed transducer and the black line shows the Aftershokz transducer. (C) The second-order and third-order distortion levels for the Aftershokz transducer when supplied with 0.1, 1, and 5 volts.

The output forces as measured on the artificial mastoid for the two types of transducers are shown in Figure 2B as force level (dB re 1 µN) in relation to 1 volt electrical input. The mastoid-placed transducer shows a classical BC transducer curve with an increase of the output with frequency at the low frequencies until a resonance (here 470 Hz) and above that, a decrease with frequency. The Aftershokz transducer shows a different type of response with an approximately flat output at frequencies below 2 kHz, above which it declines with frequency up to 10 kHz. At the highest frequencies, the Aftershokz transducer has a resonance resulting in an output peak at around 12 kHz.

It is well known that the Radioear B81 transducer generate distortions at low frequencies and high levels (Eichenauer et al., 2014; Fredén-Jansson et al., 2015). The Aftershokz transducer is less well categorized, and the distortions (second-order and third-order harmonics) were analyzed for three input levels: 0.1, 1, and 5 volts. The results from this analysis are presented in Figure 2C indicating that it is only at stimulation levels of 1 volt and over at frequencies below 0.2 kHz that the distortions exceed −20 dB. The seemingly greater high-frequency distortions at 0.1 volt compared to higher stimulation levels are due to noise limiting the estimation of the distortion component. The true distortions for 0.1 Volt stimulation level are lower than those estimated in Figure 2C. The linearity of the Aftershokz transducer was assessed by comparing the Aftershokz transfer function given in Figure 2B for stimulation levels of 0.1, 1, and 5 volts. Except for a few frequencies where rotational motion affected the output force, the results were within ± 1 dB, which is the same as the normal test–retest variability for the transducers in this setup. Consequently, nonlinear response and distortion did not significantly influence the function of the Aftershokz transducer at stimulation levels between 0.1 and 5 volts in the frequency range between 0.1 and 12.5 kHz.

As shown in Figure 2A, there were differences between the mechanical impedances of the artificial mastoid and the two stimulation positions on the head. By use of silicone pads on the artificial mastoid, new impedances were achieved that aimed to replicate those from the two stimulation positions. None of the adjusted impedances matched the impedances at the skin for the entire frequency range. However, since the artificial mastoid impedance was close to the skin impedances at frequencies above 3 kHz, the matching was done at frequencies below 2 kHz. For the mastoid position, one silicone pad gave impedances that were within 3 dB from the average skin impedance at frequencies below 2.5 kHz while for the frontal position, a reasonable match was only found for frequencies between 0.3 and 1.5 kHz using another silicone pad (Figure 3A). One reason for the large deviation at high frequencies between the artificial mastoid and the silicone pads is the lower density of the silicone compared to rubber in the artificial mastoid.

Figure 3.

(A) The mechanical point impedances of the artificial mastoid, silicone pads on the artificial mastoid, and with 2 N static force at the two stimulation positions on humans. (B) The alteration in output force from the Aftershokz transducer (blue line) and the mastoid transducer (red line) when they were measured on silicone pads applied to the artificial mastoid simulating the impedances of the stimulation positions.

The output forces from the two transducers were estimated when attached to the silicone pads and the alterations of the transducers’ output force are shown in Figure 3B. The deviation of output force from the BC transducers when placed on the silicone pads was up to 20 dB compared to when attached directly on the artificial mastoid. However, these large deviations were outside the range where the impedance of the silicone pads were like the impedance of the skin. When only focusing on the frequency ranges where the impedances of the pads mimicked the skin impedances (0.2–1.5 kHz for the frontal position and 0.1–2.5 kHz for the mastoid position), the deviation in output force varies over a range of 5 dB, and is mostly within 3 dB, to that obtained on the artificial mastoid. Consequently, for these two transducers, the mismatch in impedance between the artificial mastoid and the stimulation positions did not significantly affect the stimulation force estimation.

Hearing Thresholds

The median hearing threshold levels with stimulation at the mastoid and frontal positions are presented in Figure 4. The thresholds are shown as force levels based on the transducer calibration in Figure 2B. According to this, the ipsilateral hearing thresholds for placing the BC transducer in front of the ear canal are 10 to 40 dB better than that for a placement at the mastoid (Figure 4A). With contralateral stimulation, the median hearing thresholds for the two stimulation positions are more alike (Figure 4B) where most thresholds are within 10 dB for the two positions. The contralateral hearing thresholds are overall worse compared to the ipsilateral thresholds where the greatest difference is for frontal stimulation. The large difference in threshold levels with ipsilateral stimulation (Figure 4A) suggests different transmission pathways for the two stimulation positions. This is different from the contralateral side where the similarity in thresholds (Figure 4B) indicates similar pathways for the two stimulation positions.

Figure 4.

(A) Ipsilateral and (B) contralateral hearing thresholds in force levels when the stimulation is at the frontal position (blue line) and at the mastoid (black line). The vertical bars indicate ± 1SD and are slightly shifted for visibility. The thresholds are obtained at one-third octave frequencies while the SD bars are only presented at octave frequencies.

Simulated Cochlear Promontory Vibration

The simulated cochlear promontory accelerations from the LiUHead for frontal and mastoid stimulation (Figure 1) are shown in Figure 5. When the stimulation is a dynamic force of 1 N (Figure 5A), the response acceleration at the cochlear promontory is similar for ipsilateral and contralateral excitation at frequencies below 2 kHz. However, the response acceleration at the cochlear promontory differs between frontal and mastoid stimulation positions. The frontal stimulation gives 10 to 15 dB higher promontory acceleration than stimulation at the mastoid for frequencies below 600 Hz. At frequencies of 1 kHz and above, stimulation applied at the mastoid gives 5 to 20 dB higher cochlear promontory vibration response compared to applying the stimulation at the frontal position. Also, at frequencies of 2 kHz and above, there is a difference in cochlear promontory response vibration for ipsilateral and contralateral applied stimulation that increases with frequency, amounting to 20 dB at 10 kHz.

Figure 5.

(A) The simulated acceleration at the cochlear promontory when the stimulation was a dynamic force of 1 N applied at the frontal position (blue line: ipsilateral, red line: contralateral) and at the mastoid (black line: ipsilateral, magenta line: contralateral). (B) The simulated acceleration at the cochlear promontory when the stimulation was the threshold force from Figure 4.

When the simulated cochlear promontory vibrations are related to the force levels at threshold in Figure 4, the outcomes change (Figure 5B). In theory, if the stimulations lead to BC hearing dominated by skull bone vibrations, the simulated cochlear promontory accelerations in Figure 5B should be independent of stimulation position (frontal, mastoid, ipsilateral, or contralateral). However, since hearing thresholds are associated with some uncertainty and the vibration analysis is obtained in a mathematical model with its geometry based on one individual, the results are not expected to be identical. If a deviation of 10 dB is acceptable, the three positions ipsilateral mastoid, contralateral mastoid, and contralateral frontal produce similar cochlear promontory vibration at threshold for frequencies between 450 Hz and 3 kHz. At higher frequencies, only stimulations at the mastoids give similar cochlear vibrations, and at low frequencies, only the contralateral stimulations give similar cochlear promontory vibrations.

An interpretation of these results is that skull bone vibrations dominate the response for BC stimulation at the mastoid and the frontal position on the contralateral side. But for frontal stimulation on the ipsilateral side, a different pathway dominates the sound perception. The deviation between the ipsilateral mastoid and the contralateral stimulations at lower frequencies can be a result of the LiUHead lacking a neck connection. The lack of a neck connection seems to improve the ipsilateral response 5 to 10 dB more at frequencies below 1 kHz compared to the contralateral response (Prodanovic & Stenfelt, 2021). The simulated difference at low frequencies in Figure 5B is therefore also likely affected by the lack of neck connection. The high-frequency difference between the responses from the frontal stimulation on the contralateral side and the stimulation on the mastoid positions in Figure 5B has a different explanation. When investigating the hearing threshold force levels in Figure 4, the difference between ipsilateral and contralateral force levels for frontal stimulation is around 30 dB at frequencies up to 1 kHz, above which it increases with frequency up to 40 dB at 3 kHz. When testing hearing thresholds, the masking level of the nontest ear was set at 30 dB above the level of the test tone. However, when the interaural level difference becomes larger than 30 dB at higher frequencies, as shown in other studies (Stenfelt, 2012), the level of the ipsilateral BC stimulus needed to stimulate the contralateral ear surpasses the level of the masking noise in the ipsilateral ear, allowing a response from the ipsilateral ear at levels below the contralateral threshold. As a result, the hearing testing with frontal stimulation on the contralateral side can be biased at frequencies above 3 kHz.

Ear Canal Sound Pressure

The details of the ear canal sound pressure with AC and BC stimulations at the mastoid are provided in Surendran and Stenfelt (2022), and the focus of the current study is the results with BC stimulation on the frontal position. Figure 6 shows the ear canal sound pressure in the open ear when the stimulation is a 1 N dynamic force at the frontal position with both ipsilateral and contralateral applications. The individual data of ear canal sound pressure with frontal stimulation show that, except for a few outliers and antiresonances at isolated frequencies, the range of the individual results is approximately 20 dB (Figure 6). At the ipsilateral side, for frontal stimulation, the median ear canal sound pressure is between 85 and 105 dB SPL at frequencies up to 2 kHz, above which it becomes near 110 dB SPL up to 10 kHz. For the same stimulation position, the contralateral ear canal sound pressure is lower than the ipsilateral stimulated ear canal sound pressure with median values of 50 to 70 dB SPL at frequencies below 1 kHz and 70 to 80 dB SPL at frequencies between 1 to 10 kHz. Consequently, there is a 30 to 40 dB difference between the ipsilateral and contralateral ear canal sound pressures with frontal BC stimulation.

Figure 6.

(A) Ipsilateral and (B) contralateral ear canal sound pressure in the open ear when the frontal stimulation is a dynamic force of 1 N. The individual data are shown in black thin lines and the medians are shown as thick magenta lines. Also, the median ear canal sound pressure levels with 1 N stimulation at the mastoid are included and shown as a thick red line.

With ipsilateral stimulation, the median ear canal sound pressure in the open ear is 20 to 40 dB lower when the stimulation is applied to the mastoid compared to the frontal position. This indicates that the frontal stimulation is more efficient in generating a sound pressure in the ipsilateral ear canal than the mastoid stimulation. This changes when the stimulation is at the contralateral side (Figure 6B). Here, the two stimulation positions are relatively equal in generating an ear canal sound pressure that mainly fall within 10 dB.

The ear canal sound pressure was also measured when the ear canal was occluded, which enables an analysis of the occlusion effect with stimulation at the four stimulation positions. The occlusion effect is here defined as the ear canal sound pressure measured with the earplug in relation to the ear canal sound pressure with the ear canal open. The median occlusion effects for the four stimulation positions used are shown in Figure 7A. There is a general agreement for all four positions with a positive occlusion effect at the low frequencies that increases with decreasing frequency, and a negative occlusion effect at frequencies above 2 kHz. But there are also differences. The ipsilateral frontal stimulation generates the lowest occlusion effects overall and for frequencies below 1.5 kHz, this occlusion effect is around 5 dB below the occlusion effect with ipsilateral mastoid stimulation. The low-frequency occlusion effects for stimulation at the contralateral positions and the ipsilateral mastoid position vary with frequency in an irregular fashion but show mainly the same occlusion effect at frequencies below 1.5 kHz. These results are here interpreted as that the stimulation close to the ear canal (frontal position), forces the cartilage and soft tissues around the ear canal to vibrate more compared to a stimulation further away (mastoid, contralateral side). The positions further away involve more vibration of the bony part of the ear canal. This larger ear canal sound pressure generation with frontal stimulation is supported by the results in Figure 6A. When the earplug is introduced into the ear canal, it removes a large part of the sound radiated from the soft tissue part of the ear canal. Thereby it affects the occluded ear canal sound pressure relatively more when the stimulation is close to the ear canal than when it is further away.

Figure 7.

(A) The median occlusion effects with stimulation at the two ipsilateral positions and at the two contralateral positions. The occlusion was achieved by a foam earplug placed approximately 15 mm into the ear canal. (B) The sound pressure level difference between the sound pressure radiated from the Aftershokz transducer at the artificial mastoid and the ear canal sound pressure with the Aftershokz transducer at the frontal position (blue line). The red line shows the same difference when corrected for the effect of the ear canal according to Shaw (1974).

Another interpretation of the results in Figure 7A is that the Aftershokz transducer radiates airborne sound that is transmitted to the ear canal in the open ear thereby reducing the occlusion effect. The transducer's sound radiation was investigated by estimating the airborne sound in relation to the ear canal sound pressure similar to the analysis in Surendran and Stenfelt (2022). The airborne sound from the Aftershokz transducer was measured with the transducer on the artificial mastoid and the probe tube opening of the ER7C microphone at the approximate position of the ear canal opening. This sound pressure was compared with the ipsilateral open ear when the Aftershokz transducer is placed in its normal position (Figure 6A). The comparison (Figure 7B, blue line) show that the radiated sound pressure is 20 to 30 dB below the ear canal sound pressure at frequencies up to 8 kHz; it is only at 10 kHz that the radiated sound becomes close to the ear canal sound pressure with a difference of only 5 dB. To account for the effect of ear canal on the radiated sound, data from Figure 2 in Shaw (1974) were used to adjust the radiated sound from the entrance of the ear canal to the eardrum (red line in Figure 7B). This adjustment did not change the general outcome and the analysis revealed that the radiated sound from the Aftershokz transducer itself is not the main source of the ear canal sound pressure in the open ear even though its position is close to the ear canal opening.

Ear Canal Sound Pressure at Threshold

The measurement of hearing thresholds (Figure 4) and ear canal sound pressure (Figure 6) facilitates a comparison of the ear canal sound pressure at threshold for the different stimulation positions and modalities. Figure 8 shows the medians and standard deviations for the ear canal sound pressure at threshold when the stimulation is by AC, and by BC at the frontal and mastoid positions on the ipsilateral and contralateral sides. The AC ear canal sound pressure at threshold is considered the lowest ear canal sound pressure level required to evoke a hearing perception. However, at most frequencies with both ipsilateral and contralateral stimulation, the BC generated ear canal sound pressures at threshold exceeds the AC generated ear canal sound pressure. As explained in Surendran and Stenfelt (2022), this is consistent with the use of masking in the BC threshold testing. The use of masking elevates the hearing thresholds and it was estimated in Surendran and Stenfelt (2022) to raise the BC threshold levels by 5 to 10 dB compared to the AC threshold levels.

Figure 8.

The median sound pressure level at threshold when the BC stimulations are at the (A) ipsilateral and (B) contralateral side. The AC stimulation (red curve) is at the ipsilateral side for both panels. The vertical bars indicate ± 1SD and are slightly shifted for visibility. AC: air conduction; BC: bone conduction.

At frequencies up to 4 kHz, the ipsilateral BC generated ear canal sound pressures at threshold are similar and are 0 to 10 dB higher than the AC generated ear canal sound pressures at threshold (Figure 8A). It should be noted that the standard deviations of the sound pressures are 5 to 15 dB and the differences between the medians of the three stimulations are mostly less than the standard deviations at the same frequency. At frequencies above 4 kHz, there is a difference in the median ipsilateral sound pressures between mastoid and frontal stimulation: the mastoid stimulation gives down to 15 dB lower ear canal sound pressure levels than frontal stimulation at threshold. Also, at the highest frequency, the median AC ear canal sound pressure at threshold is 8 dB higher than that from BC frontal stimulation. However, at high frequencies, there is an uncertainty in the measurement of ear canal sound pressure due to standing waves in the ear canal, and the placement of the probe tube microphone within 5 mm from the eardrum can result in up to 7 dB variation in the sound pressure measurement at 12.5 kHz (Surendran & Stenfelt, 2022).

The results with contralateral stimulation in Figure 8B show similar results as with ipsilateral stimulation. At low frequencies, the frontal stimulation gives results that are more similar to the AC data than the BC mastoid data, but the median results of the two contralateral BC stimulations are within a few dBs at frequencies between 0.8 and 6 kHz. At the highest frequency, the median sound pressure at threshold for BC mastoid stimulation is around 10 dB below BC frontal stimulation and 35 dB below AC stimulation.

To investigate the results in Figure 8 further, a repeated measures analysis of variance (ANOVA) was computed for the frequencies where thresholds for all stimulations were measured (250, 500, 1 k, 2 k, 4 k, 8 k, 12.5 k Hz). The ANOVA included two within-subject factors (Stimulation, Frequency) where Frequency violated Mauchly's test of sphericity (χ²(20) = 70.18, p < .001) and the degrees of freedom were adjusted according to Greenhouse-Geisser. With these adjustments, the ANOVA showed a significant effect of stimulation [F(4,84) = 7.17, p < .001] and frequency [F(2.85,59.78) = 6.57, p < .001], and a significant interaction between stimulation and frequency [F(10.28,215.97) = 12.87, p < .001]. The significance of stimulation was driven by the ipsilateral frontal stimulation that gave significantly higher ear canal sound pressures at threshold compared to the contralateral stimulations (p = .042 mastoid, p < .001 frontal position). No other comparison of the stimulations was significant even if the difference between AC stimulation and contralateral frontal stimulation approached significance (p = .055). The significance of Frequency was mainly driven by the sound pressures at frequencies 0.5 and 1 kHz being lower than at 250 Hz, 2 kHz, and 12.5 kHz (p ranging between .006 and .045). The interaction showed that the differences between the stimulations were mainly significant at the two highest frequencies (8 and 12.5 kHz) where the AC results differed from the BC results, the BC ipsilateral and contralateral results differed, and the ipsilateral BC frontal differed to the BC mastoid positions.

To visualize the results further, the median and standard deviation of the differences between the BC and the AC stimulations are displayed in Figure 9. Here, the similarities between the two BC ipsilateral stimulation positions at frequencies up to 4 kHz become clear as well as the similarities between the two contralateral stimulation positions. The difference between AC and BC stimulations at frequencies above 4 kHz for all positions, except with ipsilateral BC frontal stimulation, is also visible. These data strengthen the conclusion from Surendran and Stenfelt (2022) that the ear canal sound pressure with BC stimulation in an open ear canal gives, at least, a similar contribution to BC hearing as other pathways at frequencies between 0.25 and 4 kHz. For ipsilateral BC stimulation close to the ear canal opening, this conclusion extends to higher frequencies.

Figure 9.

The difference in ear canal sound pressure at threshold (in dB) between BC and AC stimulation when the BC stimulation is at (A) the ipsilateral and (B) the contralateral side. The vertical bars indicate ± 1SD and are slightly shifted for visibility. AC: air conduction; BC: bone conduction.

As already explained, masking of the nontest ear in BC threshold testing can elevate the hearing thresholds due to central masking, which is not observed in AC testing. Surendran and Stenfelt (2022) previously assumed that this effect could range from 5 to 10 dB. This study aims to investigate this effect further. Our findings indicate that the dominant pathway for BC hearing is through the ear canal sound pressure when the stimulation is at the frontal position (Figure 5B). This means that at threshold, the ear canal sound pressure should be the same for AC stimulation as for BC frontal stimulation. By comparing these two results, the difference between them can provide an estimate of the central masking effect on threshold testing resulting from masking the nontest ear.

We investigate the possible magnitude of this central-masking effect using the comparison of ear canal sound pressure at AC and BC thresholds from Figure 9. To avoid uncertainties in the ear canal sound pressure measurement caused by standing waves, the analysis is limited to frequencies of 4 kHz and below. Also, since the results from stimulation at the mastoid are nearly identical to the results from frontal stimulation in Figure 9A at frequencies below 4 kHz, the same analysis is done for both stimulation positions. The BC–AC SPL differences are pooled and a one-sample t-test is conducted on the data to investigate if the result is significantly different from zero. This analysis showed that the stimulation at the frontal position gave on average 2.9 dB higher sound pressure levels (p = .002, 95% confidence interval [CI]: 1.1–4.7 dB) and that the mastoid stimulation gave on average 3.8 dB higher sound pressure levels (p = .001, 95% CI: 1.5–6.1 dB) than with AC stimulation. Consequently, the current analysis indicates that the central-masking effect is in the 3 to 4 dB range, which is lower than the 5 to 10 dB range that was assumed in Surendran and Stenfelt (2022).

Transcranial Transmission

The transcranial transmission is shown in Figure 10, that is, the amount of BC sound that is transmitted to the contralateral side in relation to the ipsilateral side, for stimulation at the mastoid and frontal positions. The transcranial transmission is here computed in three ways, (1) as the difference in hearing thresholds based on the results in Figure 4, (2) as the difference in the ear canal sound pressures based on the results in Figure 6, and (3) as the difference in simulated cochlear promontory vibration levels in Figure 5A.

Figure 10.

Estimations of the BC generated transcranial transmission for the two stimulation positions. The estimations are based on force thresholds (blue and black curves), ECSP (red and magenta curves), and CP vibration from simulation in the LiUHead (cyan and green curves). BC: bone conduction; CP: cochlear promontory; ECSP: ear canal sound pressures.

Figure 10 demonstrates significant (20 dB or more) stimulus-based differences in transcranial threshold and ear canal sound pressure differences. However, within each stimulus condition, the transcranial sound pressure and threshold differences are similar. With BC stimulation at the mastoid, the transcranial transmission based on both sound pressure and hearing thresholds is close to −10 dB at 250 Hz. This increases with frequency to −3 dB at frequencies around 1 kHz, while decreasing down to −21 dB at 8 kHz and finally recovering slightly to a level of approximately −12 dB at 12.5 kHz. With frontal stimulation, the transcranial transmission starts at approximately −33 dB at 250 Hz, mainly increasing to levels between −30 and −25 dB at frequencies between 1 and 2 kHz. It then falls to −37 dB at frequencies between 3 and 4 kHz. At frequencies above 4 kHz the two curves segregate where the threshold data increase to −24 dB at 12.5 kHz while the ear canal sound pressure reaches a minima of −41 dB at 7 kHz and then increases to −34 dB at 12.5 kHz. This high-frequency separation can be explained by insufficient contralateral masking during threshold testing.

The simulated cochlear promontory vibration gives estimated transcranial transmission that is bounded by ±2 dB at frequencies below 1 kHz and is different from the others. At frequencies above 1 kHz, the simulated cochlear promontory vibration suggests a transcranial transmission that is within a couple of dBs from the estimates with mastoid stimulation. Consequently, while the simulated cochlear promontory vibration estimates a low-frequency transcranial transmission close to 0 dB, both threshold and ear canal sound pressure data suggest a transcranial transmission of between −10 and −3 dB when the stimulation is at the mastoid and −35 to −25 dB when the stimulation is at the frontal position. At higher frequencies, the simulated cochlear promontory vibration suggests a transcranial transmission that decreases with frequency similar to that with stimulation at the mastoid but around 20 dB higher than the transcranial transmission seen with frontal stimulation.

Estimation of the Ear Canal Pathway With BC Headset

The data acquired from stimulation at the mastoid and frontal positions can be used to estimate the contribution of the ear canal sound pressure to BC hearing with frontal stimulation. Three different methods are used for this estimation.

The first method assumes that with BC stimulation at the mastoid, the cochlear promontory vibration is the dominant or equal contributor to BC hearing. For BC frontal stimulation, the ear canal sound pressure is assumed to be the dominant or equal contributor to BC hearing. To estimate the contribution of ear canal sound pressure to BC hearing with frontal stimulation, the difference is computed between the cochlear promontory vibration level at threshold for mastoid position (cochlear promontory vibration driven) and frontal position (ear canal sound pressure driven). This estimate is the difference between ipsilateral frontal and mastoid curves in Figure 5B and shown in Figure 11.

Figure 11.

Estimations of the contribution from the ECSP to BC hearing with frontal stimulation. The blue curve is the difference in CP vibration at threshold for the two positions obtained from simulations in the LiUHead, the red curve is based on the force threshold-derived transcranial transmission, and the black curve is based on the ECSP-derived transcranial transmission. BC: bone conduction; CP: cochlear promontory; ECSP: ear canal sound pressures.

The other two methods use contralateral stimulation, either at the mastoid or the frontal position, assuming that BC hearing is dominated by the cochlear promontory vibration for contralateral stimulation. To account for the difference in BC sensitivity between ipsilateral and contralateral stimulation, the transcranial transmission for mastoid stimulation presented in Figure 10 is used. If the frontal stimulation at the ipsilateral side results in better (lower) thresholds than those estimated by the combination of contralateral frontal thresholds and the transcranial transmission with mastoid stimulation, these better thresholds originate from the contribution of higher levels of ear canal sound pressure. Thus, the ear canal pathway estimate is computed as the transcranial transmission for mastoid stimulation (in dB) minus the transcranial transmission for frontal stimulation. Since these estimates are obtained for both hearing thresholds and ear canal sound pressure in Figure 10, both modalities are used in Figure 11 to estimate the contribution from the ear canal sound pressure to BC hearing when the stimulation is at the frontal position.

The results in Figure 11 show that the contribution of ear canal sound pressure to BC sound perception is relatively flat for stimulation close to the ear canal opening. At most frequencies and for most estimates, the level of the ear canal contribution is between 20 and 30 dB. This means that the sound in the ear canal is 20 to 30 dB higher than other contributors for BC sound when the stimulation is close to the ear canal, such as with the Aftershokz headset used in this study. The estimates deviate the most at the highest frequencies, especially between the estimate based on the cochlear promontory vibration level at threshold and transcranial transmission from thresholds, with differences of 15 to 25 dB.

The high estimate of cochlear promontory vibration method at frequencies above 7 kHz (blue line in Figure 11) may be due to the parameters of the soft tissue in the LiUHead. In order to simulate a static force of 5.4 N for mastoid stimulation, adjustments were made to the soft tissue parameters. However, no adjustments were made for the frontal position due to the lack of experimental data available at the time. The increased stiffness of the soft tissue caused by the static force improved high-frequency vibration transmission, resulting in worse high-frequency frontal transmission compared to mastoid transmission. This led to an overestimate of the difference in cochlear promontory vibration for the two positions.

As for the transcranial transmission method based on hearing thresholds (red line in Figure 11), the likely reason for the low estimate at high frequencies is the effect of insufficient masking for contralateral BC threshold testing, as previously explained.

Discussion

The aim of the current study was to investigate BC hearing when the stimulation was at a position in front of the ear canal opening. This stimulation position is different from the classical mastoid position but has recently become popular to use for BC headsets. However, it is unclear if hearing BC sound is via the same mechanism for stimulation at the mastoid as compared to a frontal position. Here we reported significant differences in the BC sound transmission to the inner ear between frontal stimulation and mastoid stimulation. With stimulation at the mastoid, the perceived sound is related to the vibration of the bone encapsulating the inner ear (Stenfelt, 2020), while stimulation at the frontal position results in dominance of the ear canal sound pressure for BC perception in a healthy ear.

Mechanical Point Impedance

One difference between the mastoid and frontal positions is the mechanical impedance of the two positions. According to the measurements of mechanical impedance done here, the impedance for the frontal position is around 10 dB lower than at the mastoid for frequencies below 1 kHz (Figure 2A). In a recent study, Nie et al. (2022) investigated the mechanical point impedance at the mastoid and at the condyle (close to the frontal position here) with a static force of 5.4 N and a circular interface area of 175 mm². They report the average impedance level at the condyle to be around 10 dB lower than those at the mastoid at frequencies below 1 kHz, in line with the results in the present study. However, their impedances are around 5 dB lower overall at 1 kHz compared to the current results. The current results at the mastoid are within 2 dB from the impedances of the mastoid with 2 N static force of adults in Mackey et al. (2016). Moreover, Mackey et al. (2016) showed that the mastoid mechanical impedance with 5.4 N static force was 3 to 5 dB higher than with 2 N static force in adults. Consequently, the slightly lower impedances for the mastoid reported here compared to Flottorp and Solberg (1976) and Håkansson et al. (1986) is likely an effect of the lower static force used in the current study.

The mechanical impedances’ influence on the transducers’ output force was measured and shown in Figure 3. Those measurements illustrated that these two transducers were relatively insensitive to the alteration of impedance between the position on the skin and on the artificial mastoid. The greatest effect was found close to the resonances of the transducers, where the shift in resonance frequency gave a 5 to 10 dB alteration of the output. Surendran and Stenfelt (2022) made model predictions of these alterations for the transducer used at the mastoid, and their prediction of a small increase at low frequencies and a reduction of the output at higher frequencies correspond well with those measured in Figure 3B.

The Ear Canal Pathway With BC Application in Front of the Ear Canal

A stimulation close to the ear canal opening generates higher ear canal sound pressure than when the BC application is further away from the ear canal. This is apparent in Figure 6 where the ear canal sound pressure is 20 to 40 dB higher with frontal application compared to mastoid application with the same dynamic stimulation force. The source of this relatively higher sound pressure could be vibrations of the ear canal walls, but also caused by sound radiated into the air from the BC transducer itself. This latter effect was investigated in Figure 7B where it was found that the airborne sound from the Aftershokz transducer was 10 to 30 dB below the ear canal sound pressure at frequencies up to 8 kHz and around 5 dB below at frequencies of 10 kHz and above. This suggest that the sound in the ear canal is not airborne radiation from the transducer.

Another indication of the sound radiation from the transducer can be investigated in the measurement of the occlusion effect (Figure 7). The occlusion hinders sound from entering the ear canal and if the sound radiated into the surrounding air significantly influences the ear canal sound pressure with an open ear canal, the occlusion effect is reduced. Comparing the measured occlusion effects in Figure 7 shows that the occlusion effect with frontal stimulation results in down to 5 dB lower occlusion effects than with the other stimulation positions. This difference can be caused by the relatively larger vibrations of the soft tissues and cartilages surrounding the ear canal with a stimulation position close to the ear canal. Such larger vibrations induce higher ear canal sound pressure in an open ear canal, which is reduced by the insertion of an earplug in the soft tissue part of the ear canal. Reinfeldt et al. (2013) reported that the occlusion effect was significantly larger at low frequencies for ipsilateral compared to contralateral BC stimulation, indicating higher sound generation from the soft tissues for ipsilateral compared to contralateral stimulation.

It is here suggested that in a normal ear, a stimulation at the frontal position is 20 to 30 dB more sensitive than a stimulation on the mastoid (as measured by the threshold force levels, Figure 4). Nishimura et al. (2014) reported hearing thresholds in force levels to be 30 to 50 dB better when a specially built BC transducer was placed on the concha of the ear compared to placing it on the mastoid. Even if that study used a different stimulation position and different BC transducers compared to the current study, it indicates that placing the BC transducer at or very close to the ear canal improves the hearing sensitivity in a normal ear compared to placing it further away from the ear canal opening. There are studies that have investigated the hearing sensitivity with a BC transducer placement on the condyle in front of the ear canal opening and on the mastoid, similar to the positions here (Dobrev et al., 2016; Nishimura et al., 2014). Both these studies report threshold improvements in the 0 to 10 dB range for placement at the condyle compared to at the mastoid, which differ from the results here. However, the condyle is slightly anterior to the ear canal opening compared to the frontal position used in the current study. Hosoi et al. (2019) showed that the ear canal sound pressure was 30 to 40 dB higher when a BC transducer was positioned on the entrance to the ear canal compared to on the condyle. Consequently, the BC stimulation position needs to be very close to the ear canal to induce the higher sound pressure in the ear canal and improve the hearing sensitivity compared to a position further away, such as on the mastoid.

The improved hearing sensitivity with the BC transducer placed close to the ear canal opening originates in the increased ear canal sound pressure. This also means that for this position, the sound pressure in the ear canal dominates the perception of BC sound (Figure 11). For most BC stimulation positions that have been studied, the vibration of the skull bone around the inner ear determines the BC sensitivity (Stenfelt, 2016, 2020; Stenfelt & Prodanovic, 2022). This changes when the BC transducer is close to the ear canal as in the current study or is on the pinna or concha as in Nishimura et al. (2015). This means that for such positions, the sound transmission to the inner ear is similar to AC sound in a healthy ear where the ear canal sound pressure determines the hearing. This also means that the status of the outer and middle ear affects the hearing sensitivity, and this position is not appropriate for clinical use where BC and AC thresholds are used to estimate a conductive hearing loss, since both are affected by the conductive component.

BC stimulation on the pinna or concha at the ear canal opening has been suggested to result in cartilage conduction (Nishimura et al., 2014, 2015). However, as shown here, it is not different from classical BC transmission, but the ear canal component is increased and dominates in a healthy ear. In a pathological ear, for example, an ear suffering from atresia, the excitation of the inner ear originates in the bony vibrations as there is no direct cartilage path to the inner ear, and the hearing process is the same as in classic BC theory (Stenfelt, 2011; Stenfelt & Goode, 2005a). Consequently, the terminology ‘cartilage conduction’ is misleading as the sound is not transmitted uniquely by the cartilage to the inner ear. We, therefore, suggest avoiding the use of such terminology as it results in confusion about the mechanisms for hearing the sound.

Skull Bone Vibrations

The vibrations of the cochlear promontory surrounding the inner ear with stimulation at the two positions were evaluated in the LiUHead (Figure 5). These data indicate that with the same stimulation force, applying the stimulation in front of the ear canal results in 10 to 20 dB higher vibration levels compared to stimulation at the mastoid for frequencies below 0.7 kHz. One reason for this difference is the antiresonance in the BC transmission that appears at around 0.5 kHz for stimulation at the mastoid (Eeg-Olofsson et al., 2008; Stenfelt & Goode, 2005b). At higher frequencies, mastoid-applied stimulation gives higher vibration levels than frontal stimulation with a difference of nearly 20 dB at 10 kHz. This high-frequency difference between the two stimulation positions originates in the thicker soft tissue part in front of the ear (approximately 25 mm in the LiUHead) compared to at the mastoid (approximately 11 mm in the LiUHead) (Figure 1). This leads to more high-frequency attenuation for the frontal position compared to the mastoid position. Dobrev et al. (2016) found that in cadaver heads, skin-applied stimulation at the condyle gave higher high-frequency cochlear promontory vibration than stimulation at the mastoid. Eeg-Olofsson et al. (2008), also based on human cadavers, reported higher cochlear promontory vibration with direct bone stimulation at the zygomatic root (close to the frontal position here) compared to a mastoid application. Consequently, the results here are opposite to those reported in experimental studies on cadaver heads. There are several reasons for these results. In the cadaver experiments, a single-dimension laser beam was used for the vibration measurement while the LiUHead used the response from all three space-dimension to estimate the velocity. It has been reported that the cochlear promontory vibration is dominated by the stimulation direction when the stimulation is close to the cochlea while the response is in all three space dimensions without any dominating direction when the stimulation is further away (Stenfelt & Goode, 2005b). Therefore, stimulation close to the ear canal gives more vibration in line with the laser beam in the cadaver-head measurements than stimulations further away (as at the mastoid) resulting in a higher velocity measure for the stimulation close to the ear canal as compared to at the mastoid.

Another reason for the difference between the cadaver experiments and the LiUHead simulations is that in the cadavers, the static force compresses the skin resulting in higher impedance and less thickness. No compression of the soft tissues was done for the LiUHead and the original thicknesses of the soft tissues were used for both the mastoid and frontal positions. As explained previously, the soft tissue at the frontal position is thicker than at the mastoid. A thicker soft tissue leads to higher compliance, as seen in the low-frequency part of the impedance (Figure 2). The higher compliance attenuates the high-frequency sound transmission leading to the results seen in Figure 5.

Figure 11 indicates that the ear canal sound pressure dominates the sound transmission to the inner ear when the stimulation is applied at the frontal position. If the ear has a pathology that restricts the sound transmission through the outer and middle ear, the BC transmission relies in bony vibrations of the skull bone (Stenfelt, 2015). The estimation in Figure 11 of the ear canal component for BC sound applied in front of the ear canal is 20 to 30 dB higher than other components of BC hearing. Consequently, the sensitivity for application of a BC transducer at the frontal position is 20 to 30 dB lower in an ear with conductive loss compared to a healthy ear. However, the hearing thresholds for a frontal placement is 15 to 35 dB better than that for a placement at the mastoid (Figure 4). This means that the BC sensitivity from the contribution of skull bone vibrations alone is nearly the same at the two positions, which is also suggested by the cochlear promontory simulations in Figure 5. This finding is corroborated by the results in Nishimura et al. (2021) where ears suffering from aural atresia had a similar threshold independent of whether the transducer was applied at the mastoid or at the ear canal opening.

Transcranial Transmission

One nature of BC sound is that stimulation at one side of the head is transmitted to both cochleae (Eeg-Olofsson et al., 2011; Stenfelt & Goode, 2005b). This is usually known as transcranial transmission (Figure 10) or, the opposite, transcranial attenuation. The difference between the two cochleae with stimulation at the mastoid has been reported to be close to 0 dB at frequencies up to 1 kHz and increasing with frequency above this frequency (Reinfeldt et al., 2013; Stenfelt, 2012). This is in-line with the results in this study when the stimulation is at the mastoid. However, with BC stimulation at the frontal position, the transcranial transmission is −40 to −25 dB (Figure 10).

One prerequisite for binaural hearing, or spatial perception, is that the audio inputs to the two ears can be separated. For BC sound, this has been problematic, and part of the reduced binaural processing with BC applied sound is related to the cross-hearing of BC sound (Häusler et al., 1983; Mcleod & Culling, 2019; Stenfelt & Zeitooni, 2013; Zeitooni et al., 2016). With frontal application of BC sound, the interaural separation is still 10 to 30 dB worse than that estimated for earphones (Zwislocki, 1953). However, this separation is expected to provide enough isolation to enable binaural cues required for spatial perception since the transcranial transmission is worse than the effect of the head shadow (Figure 10). This indicates that BC headsets with the transducers close to the ear canal's tragus facilitate binaural hearing and are appropriate for virtual and augmented reality applications. However, to utilize this binaural stimulation, the listener cannot suffer from any conductive hearing impairment.

Conclusions

The present study demonstrated that when BC sound is applied near the ear canal, as is typically done in BC headsets, the ear canal component of the BC pathways dominates the hearing process. In a healthy ear, the sound pressure in the ear canal is 20 to 30 dB higher than in other pathways for BC hearing. This indicates that BC application near the ear canal is 15 to 35 dB more sensitive compared to placement at the mastoid, but a frontal placement is more affected by conductive hearing loss, making it less capable of differentiating sensorineural from conductive hearing losses.

Furthermore, the increase in ear canal sound pressure results in an elevated transcranial transmission loss, which falls in the −40 to −25 dB range when BC sound is applied near the ear canal. This suggests that in a healthy ear, frontal placement of BC stimulation is superior to mastoid placement in using binaural information.