The cross-sectional area of the adult ear canal differs by a factor of 2 to 3 between people at many points along its length, while a person’s left and right ears are far more alike (Balouch et al., Hearing Research 2023). That single finding explains why a custom shell made from one person’s ear geometry is a different product from a generic tip in three sizes. The canal opening is not even round: its height and width differ by about 40% (Yu et al., Applied Ergonomics 2015), and in a bench test a handheld ear scanner landed within -0.052 mm of a reference canal on average (Mallwitz et al., GMS Zeitschrift fuer Audiologie 2026). We aggregated data from CT anthropometry studies in Hearing Research, Trends in Hearing and the Journal of the Acoustical Society of America, randomized and pilot trials of ear scanning, hearing industry style data, NIOSH and Cochrane hearing protection research, and manufacturer disclosures. For the wider printer market behind these shells, see our 3D printing statistics roundup.
TL;DR
- The canal entrance is oval, with a median eccentricity of 0.64 where 0 is a circle (Keefe et al., JASA 2025).
- Canal areas differ by factors of 2 to 3 between people at many locations (Balouch et al., Hearing Research 2023).
- The canal opening’s height and width differ by about 40% (p < 0.05) (Yu et al., Applied Ergonomics 2015).
- Adult-like canal geometry emerges between ages 10 and 15, in CT scans of 221 ears (Voss et al., Trends in Hearing 2025).
- A handheld ear scanner deviated -0.052 mm on average versus -0.136 mm and -0.118 mm for silicone impressions (Mallwitz et al., 2026).
- Scanned and conventional earmolds showed no significant difference in satisfaction across 152 ears (Molhave et al., International Journal of Audiology 2026).
- All of Sonova’s custom in-the-ear hearing aid shells are 3D printed (Sonova, company statement 2019).
- Custom (and some stock) in-the-ear and in-the-canal aids were 81% of US units in 1991; all in-the-ear styles were 13% in 2020 (Hearing Review, citing HIA data, 2021).
- 1 in 5 custom hearing aids went back for shell rework in unpublished data cited in 2003 (Pirzanski and Berge, Hearing Review 2003).
- 15 of 20 novice ear impressions (75%) would have been rejected for manufacture (Pack et al., Journal of the American Academy of Audiology 2025).
- Poor earplug fit may affect 40% of users (Cochrane Database of Systematic Reviews 2024).
- A 3D printed custom earplug measured a 27 dB single-number rating versus 21 dB for foam (Forum Acusticum 2025, French-German Research Institute of Saint-Louis).
1. Ear Canal Size and Length: How Much Adults Differ
The case for custom fit starts with a number that generic tips cannot absorb: areas that differ by factors of 2 to 3 from one adult to the next (Balouch et al., Hearing Research 2023). A silicone tip offered in small, medium and large is a bet that the population clusters into three sizes. CT data says it does not.
The pattern is also uneven along the canal. The entrance is wide, the area narrows toward the eardrum, and the first bend (the first turn of the canal) is where the authors expect canal-based measurement probes to sit. Age matters too: older adults had a significantly larger first-bend area than young adults, so a shell fitted in someone’s twenties is not a lifetime fit. The size spread in the outer half of the canal, where tips and shells sit, is what decides whether a fit seals or leaks.
| Metric | Value | Source |
|---|---|---|
| Mean adult canal length (66 ears, 47 subjects aged 18 to 90) | 31.4 mm, SD 3.1 mm | Balouch et al., Hearing Research 2023 |
| Mean canal area at the entrance; at the first bend; at the eardrum end | 97 mm² (SD 24); 53 mm² (SD 18); 56 mm² (SD 8) | Balouch et al., Hearing Research 2023 |
| First-bend area, adults aged 18 to 30 vs adults aged 61 to 90 | 44 mm² vs 69 mm² (significantly larger in older adults) | Balouch et al., Hearing Research 2023 |
| Between-person difference in canal area at many locations | Factors of 2 to 3 | Balouch et al., Hearing Research 2023 |
| Median area from optical ear scans: at the entrance vs 16 mm inside | 72 mm² vs 22 mm² | Keefe et al., JASA 2025 |
| Median area from optical ear scans at 6, 9 and 12 mm inside the entrance | 47.0, 40.1 and 35.3 mm² | Keefe et al., JASA 2025 |
| Ears scanned in the 2025 CT study, and age span | 221 ears, ages 0.7 to 91 | Voss et al., Trends in Hearing 2025 |
| Area range in the middle portion of the canal (older reference cited in 2025) | 25 mm² to 70 mm² | Stinson and Lawton 1989, as cited by Voss et al., Trends in Hearing 2025 |
Note the two entrance figures: 97 mm² from CT in 2023 and 72 mm² from optical scans in 2025 (Keefe et al., JASA 2025). Where the “entrance” is drawn changes the answer, which is itself a fit problem: a lab and a scanner can disagree about where the shell should start.
2. Canal Shape: Oval Openings, Bends, Sex and Age
Size is only half the story. The height and width of the canal opening differ by about 40% (p < 0.05), which led the authors to recommend elliptical rather than circular earplug cross-sections (Yu et al., Applied Ergonomics 2015). A round foam or silicone tip can only seal an oval opening by deforming, while a shell printed from the ear’s own geometry starts out oval.
Eccentricity (a measure of how oval a cross-section is, where 0 is a perfect circle) peaks just inside the entrance and relaxes deeper in. Sex differences exist but are modest, a few square millimeters and under 2 mm of length, small next to the person-to-person spread in section 1. The more practical finding for fitting children is timing: canal geometry keeps changing until somewhere between age 10 and 15 (Voss et al., Trends in Hearing 2025), so a child’s custom shell has a short service life by anatomy alone.
| Metric | Value | Source |
|---|---|---|
| Height vs width of the canal opening | Differ by about 40% (p < 0.05) | Yu et al., Applied Ergonomics 2015 |
| Entrance diameters, largest vs smallest: men; women | 9.6 vs 6.75 mm; 9.15 vs 6.3 mm | Yu et al., Applied Ergonomics 2015, as tabulated by Keefe et al., JASA 2025 |
| Median eccentricity: at the entrance, at 4 mm inside, at 15 mm inside | 0.64, 0.74, 0.50 | Keefe et al., JASA 2025 |
| Ears in which earwax blocked CT measurements | 13 of 221 | Voss et al., Trends in Hearing 2025 |
| Tightest median bend of the canal centerline (distance measured along the canal from the entrance) | Radius of curvature 34 mm, at 8 mm | Keefe et al., JASA 2025 |
| Area where the eardrum attaches (tympanic annulus): women vs men | 61.2 vs 64.9 mm² (all 221 ears: 63.1±0.4 mm²) | Voss et al., Trends in Hearing 2025 |
| Extra canal length in men, eardrum ring to entrance (age-adjusted) | 1.8 mm | Voss et al., Trends in Hearing 2025 |
| Age by which adult-like canal features emerge | Between 10 and 15 years | Voss et al., Trends in Hearing 2025 |
Context: earwax is not only a CT problem. Any scanner or impression that has to see the canal wall stops at a wax plug, which is why custom fitting usually starts with a clean canal.
3. Impressions vs 3D Ear Scanning: Accuracy, Comfort and Workflow
For custom shells, the digital question is settled at the bench and still open in the clinic. A handheld scanner showed a mean deviation of -0.052 mm from a reference canal, versus -0.136 mm and -0.118 mm for two silicone impression materials, with a tighter spread (0.415 mm vs 0.496 to 0.546 mm) across 90 scans by five users (Mallwitz et al., GMS Zeitschrift fuer Audiologie 2026). The negative signs point to slight shrinkage in the silicone, and the authors call digital capture an equivalent alternative.
Patients like the scan more than the silicone putty: in a pilot, earmold users rated scan comfort 4.6667 out of 5 against 3.6667 for impressions (Zheng et al., Bioengineering 2023). But once the earmold is in the ear, the difference fades. A randomized trial in 76 adults found no significant difference in overall satisfaction between scanned and conventional earmolds, while 89% of screened patients were not included, a warning that not every ear can be scanned well (Molhave et al., International Journal of Audiology 2026).
| Metric | Value | Source |
|---|---|---|
| Mean deviation from a 3D printed reference canal: handheld scanner vs two silicone impression materials | -0.052 mm vs -0.136 mm and -0.118 mm | Mallwitz et al., GMS Zeitschrift fuer Audiologie 2026 |
| Standard deviation of that deviation: scanner vs impressions | 0.415 mm vs 0.496 to 0.546 mm | Mallwitz et al., GMS Zeitschrift fuer Audiologie 2026 |
| Comfort of the procedure, experienced earmold users (0 to 5 scale): scan vs impression | 4.6667 vs 3.6667 (p = 0.01) | Zheng et al., Bioengineering 2023 |
| Preference, people new to earmolds (0 to 5 scale): scan vs impression | 4.8 vs 2.4 | Zheng et al., Bioengineering 2023 |
| Training to proficiency: impressions vs scanning | About 10 hours vs 2 to 3 hours | Zheng et al., Bioengineering 2023 |
| Variable cost per earmold with scanning | About 7% lower than with impressions | Zheng et al., Bioengineering 2023 |
| Randomized trial size; satisfaction difference scanned vs conventional earmolds | 76 adults (152 ears); no significant difference | Molhave et al., International Journal of Audiology 2026 |
| Screened patients not included in that trial | 89% | Molhave et al., International Journal of Audiology 2026 |
Outlier: for hearing protection, the physical impression still wins on seal. A US Army evaluation with 20 volunteers per method found earplugs from physical impressions gave significantly higher attenuation but were the least comfortable, while plugs from three digital scanning methods did not differ from each other (Stefanson and Ahroon, JASA 2019). An earlier study found digitally scanned custom earplugs gave significantly poorer attenuation than wax-dipped custom earplugs (Gordon and Reeves, Proceedings of Meetings on Acoustics 2012). Most recent available data for the military comparison: 2019.
4. Additive Manufacturing’s Share of Custom Shells
For custom hearing aid shells, manufacturers describe the switch to 3D printing as complete. Sonova says all the shells for its custom in-the-ear hearing aids are produced on 3D printers, with digital manufacturing of custom in-the-ear aids starting in 2001/2002 (Sonova, company story 2019). A Demant production engineer writes that in-the-ear aids are almost entirely produced by additive manufacturing and have been for more than 30 years, with daily demand in excess of 1,000 pieces at his site (Demant engineer, case study on the Asiga website). Both are the companies’ own claims; no independent body publishes a printed-share figure.
The irony is that printing won just as custom styles lost the market. Behind-the-ear and receiver-in-canal (RIC, a small body behind the ear with the speaker in the canal) designs now dominate US unit sales (Hearing Review, Trends in Hearing Aid Styles 2021). The printed custom shell survives as a niche product, much like custom in-ear monitors in audio.
| Metric | Value | Source |
|---|---|---|
| Sonova custom in-the-ear shells made on 3D printers | All (vendor’s own claim) | Sonova, company story 2019 |
| Start of digital manufacturing of Sonova custom in-the-ear aids | 2001/2002 (vendor’s own claim) | Sonova, company story 2019 |
| Daily production demand at one Demant additive manufacturing site | In excess of 1,000 pieces (vendor’s own claim) | Demant engineer, Asiga case study |
| Custom (and some stock) in-the-ear and in-the-canal aids, share of US units in 1991 | 81% | Hearing Review 2021, citing HIA data |
| All in-the-ear styles, share of US units in 1996 | About 83% (6 of 7) | Hearing Review 2021, citing HIA data |
| Completely-in-canal (CIC) share of US units, 1994 vs 2001 | About 7% vs 23% | Hearing Review 2021, citing HIA data |
| All in-the-ear styles, share of US units in 2020 (half/full shell, in-the-canal, CIC) | 13% (5%, 5%, 3%) | Hearing Review 2021, citing HIA data and HR estimates |
| Receiver-in-canal share of US units in 2020 | 79% | Hearing Review 2021, citing HIA data and HR estimates |
Context: a sponsored 2017 trade article reported that Sonova owned more than 100 EnvisionTEC printers, each printing about two dozen shells per hour (3DPrint.com, sponsored by EnvisionTEC, 2017). Formlabs markets its Form 4B as able to print hearing aids or large earmolds in just a few hours (Formlabs audiology page, vendor claim). Most recent available data for US style shares: 2020.
5. Remakes, Returns and Fit Failures
Remake rates are the industry’s least published number. The best public figures are old: 1 in 5 custom hearing aids was returned to the manufacturer for shell rework in unpublished data reported in 2003 (Pirzanski and Berge, Hearing Review 2003). The same authors found that 20% to 25% of ears widened by 0.5 mm to 2.5 mm when a more viscous impression silicone or jaw movement was involved. A rigid printed shell copies the canal at one instant; a jaw that moves changes the canal it has to fit.
The technique fix was large. Earmold labs estimated that half of all hearing aid remakes come from poor impressions, and switching to open-jaw impressions with more viscous silicone remedied or reduced fitting problems in 114 of 136 patients (84%) (Pirzanski and Berge, AudiologyOnline 2002). Today’s return data covers whole devices, not shells: about 9% overall in a 2025 trial, with comfort suspected as a driver (Chen et al., Perspectives of the ASHA Special Interest Groups 2025). Our roundup of OTC hearing aid return rates covers the self-fit side of that picture.
| Metric | Value | Source |
|---|---|---|
| Custom hearing aids returned to the manufacturer for shell rework (unpublished data) | 1 in 5 | Pirzanski and Berge, Hearing Review 2003 |
| Ears that widened with a more viscous impression silicone or jaw movement (over 300 ears) | 20% to 25%, by 0.5 mm to 2.5 mm | Pirzanski and Berge, Hearing Review 2003 |
| Hearing aid remakes caused by poor impressions (earmold lab estimate) | 50% | Pirzanski and Berge, AudiologyOnline 2002 |
| Patients whose fitting problems open-jaw, higher-viscosity impressions remedied or reduced | 114 of 136 (84%) | Pirzanski and Berge, AudiologyOnline 2002 |
| Overall hearing aid return rate in a multisite randomized trial | Approximately 9% | Chen et al., Perspectives of the ASHA SIGs 2025 |
| Estimated returns: prescription vs over-the-counter hearing aids | 3% to 16% vs 30% to 50% | Chen et al., Perspectives of the ASHA SIGs 2025 |
| Novice ear impressions that would have been rejected for manufacture | 15 of 20 (75%); all 5 accepted ones would need a remake | Pack et al., Journal of the American Academy of Audiology 2025 |
Most recent available data for shell rework and impression-driven remakes: 2002 and 2003. For comparison, the 2025 trial behind the 9% figure analyzed real-ear measurements (a probe-microphone check of the sound reaching the eardrum) from 512 participants and found they did not predict who returned their aids.
6. Seal and Passive Attenuation: Custom vs Generic Fit
Hearing protection is where seal gets measured in decibels, and the lesson for custom earphones transfers directly. Poor earplug fit may be the case for 40% of users, according to the background of a 2024 Cochrane review (Morata et al., Cochrane Database of Systematic Reviews 2024). Custom molds do not automatically beat that. Their measured advantage is consistency: across repeated fittings, custom plugs gave significantly more consistent attenuation at 500, 1000 and 2000 Hz than non-custom plugs in 30 participants (Tufts et al., Annals of Occupational Hygiene 2013), and in 20 workers custom-molded plugs reached a higher share of labeled attenuation than foam on both test systems (Neitzel et al., Annals of Occupational Hygiene 2006).
A 3D printed custom plug tested on 13 volunteers in 2025 came close to its label on the single-number rating (SNR, the European one-figure summary of how many decibels a protector removes, not a signal-to-noise ratio), while foam fell short of its own (Zimpfer et al., Forum Acusticum 2025). The catch is wearability: in a NIOSH study of a semi-custom protector, only about 25% of original users were still wearing it after a year (Davis et al., Journal of Occupational and Environmental Hygiene 2011). A seal nobody wears protects nothing, and a sealed earphone shell that hurts gets left in the drawer. For the measurement side of this, see our roundup of hearing protection fit testing and PAR statistics.
| Metric | Value | Source |
|---|---|---|
| Earplug users who may have poor fit | 40% | Morata et al., Cochrane Database of Systematic Reviews 2024 |
| Single-number rating (SNR) of a 3D printed custom earplug: measured vs manufacturer | 27 dB vs 27.7 dB | Zimpfer et al., Forum Acusticum 2025 |
| SNR of a classic foam earplug: measured vs manufacturer | 21 dB vs 23 dB | Zimpfer et al., Forum Acusticum 2025 |
| Low-frequency rating (L) of the printed custom plug: measured vs manufacturer | 19 dB vs 22 dB | Zimpfer et al., Forum Acusticum 2025 |
| Custom earplug (Sonomax full-block) fit tests whose personal attenuation rating (the protection one wearer actually got) fell below the label rating after OSHA’s derating formula, before vs after training | 14% vs 9% | Murphy et al., JASA 2022 |
| Semi-custom protector: median attenuation over four test sessions; users still wearing it after a year | About 17 dB rising to about 21 dB; about 25% | Davis et al., Journal of Occupational and Environmental Hygiene 2011 |
| Mean attenuation of custom protectors from audiologist vs novice impressions | 19.51 dB vs 14.67 dB | Pack et al., Journal of the American Academy of Audiology 2025 |
| Workers who could not be protected by earplugs because of ear canal structure (50 workers tested) | 4 | Chiu and Wan, IJERPH 2020 |
Context: the 4.84 dB gap in the Pack study is our subtraction (19.51 minus 14.67) from the means it reports (Pack et al., 2025); the same 2025 study found voids in the canal portion of all 20 novice impressions. In the Chiu and Wan study, the four workers whose ear canal structure defeated the earplugs were protected only after switching to earmuffs (Chiu and Wan, IJERPH 2020). The Sonomax comparison comes from a NIOSH dataset in which some workers rejected the custom plug for low attenuation and discomfort (Murphy et al., JASA 2022). Seal also shapes how an earphone’s tuning reaches the eardrum, a theme behind our Harman target curve preference roundup.
Summary: Custom 3D Printed Headphone Ear Canal Fit by the Numbers
| Metric | Value | Source |
|---|---|---|
| Mean adult canal length | 31.4 mm, SD 3.1 mm | Balouch et al., Hearing Research 2023 |
| Between-person difference in canal area at many locations | Factors of 2 to 3 | Balouch et al., Hearing Research 2023 |
| Height vs width of the canal opening | Differ by about 40% (p < 0.05) | Yu et al., Applied Ergonomics 2015 |
| Median eccentricity at the canal entrance | 0.64 | Keefe et al., JASA 2025 |
| Age by which adult-like canal features emerge | 10 to 15 years | Voss et al., Trends in Hearing 2025 |
| Mean deviation: handheld scanner vs silicone impressions | -0.052 mm vs -0.136 and -0.118 mm | Mallwitz et al., 2026 |
| Procedure comfort, scan vs impression (0 to 5) | 4.6667 vs 3.6667 | Zheng et al., Bioengineering 2023 |
| Satisfaction, scanned vs conventional earmolds (152 ears) | No significant difference | Molhave et al., International Journal of Audiology 2026 |
| Screened patients not included in the scan trial | 89% | Molhave et al., International Journal of Audiology 2026 |
| Sonova custom in-the-ear shells made on 3D printers | All (vendor’s own claim) | Sonova 2019 |
| Custom in-the-ear and in-the-canal share of US units, 1991 vs all in-the-ear styles in 2020 | 81% vs 13% | Hearing Review 2021, citing HIA data |
| Custom hearing aids returned for shell rework | 1 in 5 | Pirzanski and Berge, Hearing Review 2003 |
| Overall hearing aid return rate in a 2025 trial | Approximately 9% | Chen et al., Perspectives of the ASHA SIGs 2025 |
| Novice impressions rejected for manufacture | 15 of 20 (75%) | Pack et al., 2025 |
| Earplug users who may have poor fit | 40% | Cochrane Database of Systematic Reviews 2024 |
| SNR measured: 3D printed custom plug vs foam plug | 27 dB vs 21 dB | Zimpfer et al., Forum Acusticum 2025 |
| Attenuation from audiologist vs novice impressions | 19.51 dB vs 14.67 dB | Pack et al., 2025 |
| Semi-custom protector users still wearing it after a year | About 25% | Davis et al., 2011 |
Methodology and Sources
- Balouch, Bekhazi, Durkee et al., Measurements of ear-canal geometry from high-resolution CT scans of human adult ears, Hearing Research 2023
- Voss, Remenschneider, Farrar, Myoung and Horton, Comprehensive measurements and analyses of ear canal geometry from late infancy through late adulthood, Trends in Hearing 2025
- Keefe, Porter and Fitzpatrick, Shape and sound analyses of the human ear-canal geometry, Journal of the Acoustical Society of America 2025
- Yu, Lee, Wang et al., Anthropometry of external auditory canal by non-contactable measurement, Applied Ergonomics 2015
- Mallwitz, Joukov, Albrecht and Kreikemeier, Digital versus conventional impression-taking of the ear canal, GMS Zeitschrift fuer Audiologie 2026
- Zheng, Didion and Ferguson, Comparison of traditional impression and 3D ear scanning techniques: a pilot study, Bioengineering 2023
- Molhave, Madsen, Degnbol, Christensen and Ovesen, Three-dimensional scan versus conventional mould of outer ear in fabricating earmolds: a randomized controlled trial, International Journal of Audiology 2026
- Stefanson and Ahroon, Evaluation of digital ear scanning for custom hearing protection devices, JASA 2019
- Gordon and Reeves, Real-ear attenuation of custom-fit earplugs with the communications earplug, Proceedings of Meetings on Acoustics 2012
- Sonova, 3D printing technology improved hearing, company story 2019
- Demant production engineer, UltraGLOSS 3D printed hearing aids, case study on the Asiga website
- Strom, Trends in hearing aid styles, Hearing Review 2021 (HIA data and Hearing Review estimates)
- 3DPrint.com, How Sonova became a global leader in production 3D printing, sponsored by EnvisionTEC, 2017
- Formlabs, 3D printing in audio, vendor page
- Pirzanski and Berge, Earmold impressions: does it matter how they are taken?, Hearing Review 2003
- Pirzanski and Berge, Ear impressions: art or science?, AudiologyOnline 2002
- Chen, Remshak, Meskan, Singh, Pitman and Dhar, Real-ear measures, APHAB and hearing aid retention, Perspectives of the ASHA Special Interest Groups 2025
- Pack, Bryan and Gordon-Hickey, Effects of consumer versus professional earmold impression-taking on custom hearing protection device attenuation, Journal of the American Academy of Audiology 2025
- Morata, Gong, Tikka, Samelli and Verbeek, Hearing protection field attenuation estimation systems and associated training, Cochrane Database of Systematic Reviews 2024
- Tufts, Jahn and Byram, Consistency of attenuation across multiple fittings of custom and non-custom earplugs, Annals of Occupational Hygiene 2013
- Neitzel, Somers and Seixas, Variability of real-world hearing protector attenuation measurements, Annals of Occupational Hygiene 2006
- Zimpfer, Hamery and Blonde-Weinmann, Comparison of REAT and insertion loss measurements using headphone audiometry, Forum Acusticum 2025
- Murphy, Gong, Karch, Federman and Schulz, Personal attenuation ratings versus derated noise reduction ratings for hearing protection devices, JASA 2022
- Davis, Murphy, Byrne and Shaw, Acceptance of a semi-custom hearing protector by manufacturing workers, Journal of Occupational and Environmental Hygiene 2011
- Chiu and Wan, Individual fit testing of hearing-protection devices based on microphones in real ears, International Journal of Environmental Research and Public Health 2020
- Data watch: no independent body publishes the share of hearing aid shells or custom in-ear monitor shells made by additive manufacturing, so section 4 relies on manufacturer statements (labeled as vendor claims) plus US style shares that end in 2020. The 2023 Hearing Research abstract prints the canal length unit as mm² (a typo for mm), and its entrance area of 97 mm² from CT compares with 72 mm² from optical scans (the 2025 JASA study, which also cites 72 mm² from Stinson and Lawton 1989). Those 2025 optical scans reached only about 8 to 18 mm into the canal and rarely reached the second bend, so scanner data under-describes the deep canal. The Zheng pilot had 12 participants and its paper contains inconsistent statistics (for example F(1,21) = 0.46 paired with p = 0.006). The 1-in-5 shell rework and 50% impression-remake figures in section 5 are from 2002 and 2003 and partly rest on unpublished or estimated data. The Chiu and Wan paper reports different totals for its test count (148, 140 and 111 Classic trials in different places), so only its count of four workers is used here. Remake and return rates for custom in-ear monitors are not published by any manufacturer we could find.
- Last updated: October 27, 2026. We update this roundup quarterly, and the next refresh is expected when the full results of the Molhave randomized trial (remake and adjustment counts) and the next Hearing Industries Association unit statistics by style are published.