57.5% of 1,522 US adults say they are dissatisfied when they listen to a recording of their own voice, compared with 38.8% who dislike how they sound in normal conversation (Naunheim et al., Laryngoscope 2023). That gap of almost 19 percentage points is the measurable core of what psychologists have called voice confrontation since the 1960s: the jolt of meeting your voice as other people hear it. The mismatch has a physical basis, because a microphone misses the bone-conducted sound you hear inside your own head, and it has a social side, since in one 2026 experiment only 59 of 139 listeners spontaneously recognized their own voice at all. Voice notes, video calls and even voice prompts on dating apps turn that confrontation into a routine event. We aggregated data from the Laryngoscope, YouGov, the Journal of the Acoustical Society of America, Cognition, Trends in Hearing, the International Journal of Psychology, Royal Society Open Science, and more than a dozen other peer-reviewed sources.
TL;DR
- 57.5% of US adults report dissatisfaction when they hear a recording of their own voice, vs 38.8% in normal conversation (Naunheim et al., Laryngoscope 2023).
- 43% of 4,058 US adults are somewhat (27%) or very (16%) uncomfortable listening to a recording of their voice (YouGov Daily Question, December 18, 2019).
- The bone-conducted part of your own voice dominates between 1 and 2 kHz for most phonemes tested (Reinfeldt et al., JASA 2010).
- Only 2 of 10 participants picked the unaltered recording as the version most like their own voice (Kimura and Yotsumoto, PLoS One 2018).
- Listeners who recognized their own voice rated it more attractive than other voices; the 80 who did not recognize it rated it less attractive (Rosi et al., Cognition 2026).
- Own-voice recognition reached 85.6% in a first language but 74.4% in a second language (Jogia et al., International Journal of Psychology 2024).
- Social anxiety correlated with disliking one’s own recorded voice at r = -.228 (Jogia et al., International Journal of Psychology 2024).
- 40% of first-time hearing aid users rated their live voice as less pleasant with hearing aids (Hengen et al., Trends in Hearing 2025).
- 50.6% of respondents without a voice disorder would consider an intervention to change their voice (Naunheim et al., Laryngoscope 2023).
- Familiar listeners spotted clones of their own voice with a d-prime of 2.13; strangers performed at chance (Rosi et al., Computers in Human Behavior: Artificial Humans 2025).
1. How Many People Dislike Their Recorded Voice
The most useful number here is not the headline 57.5% but the difference between the two questions. The same respondents who mostly accept their voice in conversation turn against it when it plays back through a speaker, which suggests the recording itself, not a general voice insecurity, does much of the damage. Recording dissatisfaction ran 18.7 percentage points higher than conversational dissatisfaction in the Laryngoscope population survey (57.5% minus 38.8%, our calculation), whose 1,522 respondents had an age, gender and regional distribution reflective of the US population.
The YouGov poll asked a softer question (comfort rather than satisfaction) and found Americans almost evenly split: 48% comfortable (23% very plus 25% somewhat) against 43% uncomfortable. Most recent available data: YouGov, 2019. The different wording likely explains part of the gap with the clinical survey, so the two figures should be read side by side, not averaged.
| Metric | Value | Source |
|---|---|---|
| Respondents who did not like the sound of their own voice in normal conversation | 38.8% | Naunheim et al., Laryngoscope 2023 |
| Respondents who reported dissatisfaction when listening to a recording of their own voice | 57.5% | Naunheim et al., Laryngoscope 2023 |
| Factors associated with voice discontent | Middle age (p = 0.005), female gender (p < 0.0001), white race (p < 0.0001) | Naunheim et al., Laryngoscope 2023 |
| US adults ‘very uncomfortable’ listening to a recording of their own voice | 16% | YouGov Daily Question, December 18, 2019 (n = 4,058) |
| US adults ‘somewhat uncomfortable’ listening to a recording of their own voice | 27% | YouGov Daily Question, December 18, 2019 |
| US adults very or somewhat uncomfortable (16% + 27%, our calculation) | 43% | YouGov Daily Question, December 18, 2019 |
| US adults ‘very comfortable’ listening to a recording of their own voice | 23% | YouGov Daily Question, December 18, 2019 |
| Survey participants who felt negative / neutral / positive about their own voice | 31% / 24% / 44% | Chong et al., Journal of Voice 2024 (n = 100), as reported by Jogia et al., International Journal of Psychology 2024 |
Context note: the Laryngoscope sample was about 52% female, with roughly 30% aged 18 to 34, 32% aged 35 to 54 and 38% aged 55 or older, according to the ENT Today summary of the paper. YouGov also publishes 9% ‘don’t know’ answers, so its comfortable and uncomfortable shares do not add to 100%.
2. Why a Recording Sounds Wrong: Bone Conduction vs Air Conduction
When you talk, your voice reaches your inner ear by two routes: through the air into the ear canal, and as vibration through the bones of the skull (bone conduction). A microphone records only the first route, so the playback is missing part of the sound you have heard your whole life. The bone-conducted part dominated between 1 and 2 kHz for most of the ten phonemes measured by Reinfeldt and colleagues at Chalmers University of Technology (Journal of the Acoustical Society of America 2010). Most recent available data: Reinfeldt et al., 2010.
The perceptual studies show why no simple fix exists. When researchers offered filtered versions of each participant’s recording, most people chose a filtered one as the closest match to how they hear themselves, but no single filter won for everyone. In Kimura and Yotsumoto the authors concluded that ‘there was no general filter that represented own voice’. A 2026 preprint goes further: voices rebuilt from participants’ judgments were brighter than a bone-conduction model predicts, which hints that the mental image of one’s own voice is partly idealized rather than purely acoustic.
| Metric | Value | Source |
|---|---|---|
| Frequency range where the bone-conducted part of one’s own voice dominated (most of 10 phonemes tested) | 1 to 2 kHz | Reinfeldt et al., JASA 2010 |
| Behavior of body-conducted own voice in an ear blocked by an earpiece (19 participants, 37 ears) | Amplified below about 700 Hz, attenuated at higher frequencies | Acta Acustica occlusion effect methods study 2025 |
| Participants who chose the unadjusted recording as most like their own voice | 2 of 10 (8 chose a modified version) | Kimura and Yotsumoto, PLoS One 2018 |
| Consistency of the ‘most own-voice-like’ choice across sessions | Spearman’s rho = .899 | Kimura and Yotsumoto, PLoS One 2018 |
| Participants who picked the raw recording as most like their own voice, soundproof room | 6 of 17 (35.3%, our calculation) | Hosaka et al., Scientific Reports 2021 |
| Participants who picked the raw recording as most like their own voice, inside an MRI scanner | 3 of 17 (17.6%, our calculation) | Hosaka et al., Scientific Reports 2021 |
| Mean absolute gap between the pitch people perceive as their own and the pitch they produce (n = 22) | 8.52 Hz (0.9 semitones) | Neamaalkassis et al., Royal Society Open Science 2025 |
| Japanese adults tested on a reconstructed self-voice (Study 1 / Study 2) | n = 24 / n = 54; reconstructed voices rated more own-voice-like than recordings or bone-conduction simulations | Suzuki et al., PsyArXiv preprint 2026 |
Context note: the small 0.9-semitone pitch gap in the Neamaalkassis study suggests that people know their basic pitch (fundamental frequency, or F0) well, so the ‘that is not me’ reaction probably comes more from timbre, the color of the sound, than from pitch. The Suzuki et al. preprint reports higher spectral centroid and F1 values in the reconstructed voice, which in plain terms means a brighter, clearer sound.
3. The Attractiveness Paradox: Liking a Voice You Do Not Recognize
For a decade the standard counterpoint to voice confrontation was a self-enhancement finding: if you do not know a clip is you, you rate it kindly. In Hughes and Harrison (Perception 2013), 80 men and women rated recordings of people counting from one to ten (the study stimulus) without being told their own voices were included, and they rated their own voice as more attractive than other raters did. The authors called this ‘vocal implicit egotism’. Most recent available data for this design: Hughes and Harrison, 2013.
The newest data complicates that story. Rosi, Lavan and McGettigan found the opposite direction for people who did not recognize themselves: the 80 non-recognizers rated their own voice as less attractive than other voices, while the 59 recognizers rated it as more attractive (Cognition 2026). The question is no longer whether people secretly like their voice, but how much awareness changes the verdict. Recognition rates themselves are covered in section 4.
| Metric | Value | Source |
|---|---|---|
| Participants who rated an array of voices that secretly included their own | 80 men and women; own voice rated more attractive than other raters rated it | Hughes and Harrison, Perception 2013 |
| Participants who spontaneously recognized their own voice during a rating task | 59 of 139 (42.4%, our calculation) | Rosi et al., Cognition 2026 |
| Direction of own-voice rating vs other voices | Recognizers: more attractive; non-recognizers: less attractive | Rosi et al., Cognition 2026 |
| Altered versions of own voice rated less attractive and less similar than the plain recording | Pitch +/-20 Hz versions (vs loudness +/-10 dB versions and the recording) | Peng et al., Australian Journal of Psychology 2019 |
| Self-enhancement in voice attractiveness ratings vs recognition | Significant whether or not participants recognized their self-voice | Peng et al., Scandinavian Journal of Psychology 2020 |
| Friend-enhancement effect (rating a friend’s voice higher) | Found only in participants who recognized the friend’s voice | Peng et al., Scandinavian Journal of Psychology 2020 |
| Attractiveness of a friend’s recording vs one’s own recording, rated by 47 familiar listeners | Friend rated higher (p < .001) | Rosi et al., Computers in Human Behavior: Artificial Humans 2025 |
Outlier note: the Peng et al. 2020 study and the Rosi et al. 2026 study disagree on what happens when people fail to recognize their voice. Different stimuli, rating scales and recognition checks are the likely reason, and Rosi et al. argue that earlier work never tested recognition rigorously. The Peng et al. 2019 experiments add one practical detail: a pitch shift of only 20 Hz was enough to make one’s own voice less attractive and less familiar than a loudness change of 10 dB.
4. Recognizing Your Own Voice on a Recording
Self-voice recognition is good but far from perfect, and it is fragile. Strip a recording down to a narrow frequency band and accuracy for self and familiar voices falls from 81.81% to about 55% (Xu et al., Frontiers in Psychology 2013). Switch the speaker into a second language and accuracy drops by double digits. Own-voice recognition fell from 85.6% in Arabic to 74.4% in English among Arabic-English bilingual women from the same university sample (Jogia et al., International Journal of Psychology 2024).
Even when recognition fails, the brain still treats one’s own voice as special. In Kirk and Cunningham (British Journal of Psychology 2025), reaction time and accuracy favored the participant’s own recorded voice even when it was assigned the label ‘friend’ or ‘stranger’. That is the self-prioritization effect: self-related cues get processed faster.
| Metric | Value | Source |
|---|---|---|
| Recognition accuracy for self and colleague voices with the full frequency band (30 Japanese adults, chance = 20%) | 81.81% | Xu et al., Frontiers in Psychology 2013 |
| Self-voice vs other-voice recognition, high-frequency band only | 60.17% vs 49.67% (F(1, 29) = 4.55, p = 0.04) | Xu et al., Frontiers in Psychology 2013 |
| Own-voice recognition, first language (Arabic) vs second language (English) | 85.6% vs 74.4% | Jogia et al., International Journal of Psychology 2024 |
| Self-voice hit rate in a control recognition task | 0.79 (95% CI 0.72-0.85) | Orepic et al., Royal Society Open Science 2023 |
| Participants excluded for recognizing their own unaltered voice in fewer than half of trials | 9 of 52 (17.3%) | Orepic et al., Royal Society Open Science 2023 |
| Reaction time and accuracy for one’s own recorded voice vs external voices, even when it carried a ‘friend’ or ‘stranger’ label | Consistent own-voice advantage (N = 90) | Kirk and Cunningham, British Journal of Psychology 2025 |
| Accuracy distinguishing one’s own voice from a stranger’s voice in a 20-trial pilot (10 participants) | 100% | Zheng et al., PLoS One 2011 |
Context note: in the Orepic et al. study, adding bone-conducted vibration to the playback improved telling self from other but not telling a familiar voice from a stranger, which the authors read as evidence that one’s own voice is ‘a fundamentally multi-modal construct’. The perfect score in the Zheng et al. pilot came from a simple two-voice check, which is why it sits so far above the 79% hit rate in harder designs. Most recent available data for the Xu and Zheng designs: 2013 and 2011.
5. Social Anxiety and Voice Confrontation
The term voice confrontation goes back to Holzman and Rousey, whose experiments in the Journal of Personality and Social Psychology in 1966 and 1967 studied how people react when they hear their recorded voice, including a 1967 bilingual study. Nearly six decades later the bilingual angle has returned, now tied to social anxiety, measured here with the Liebowitz Social Anxiety Scale (LSAS), a standard questionnaire.
The correlations are small but consistent. Higher social anxiety went with lower liking of one’s own recorded voice in both languages (r = -.228 in Arabic, r = -.181 in English) in the same bilingual sample described in section 4. The intervention data point the same way: in the Lund University experiments, socially anxious people rated their recorded speeches less harshly when the playback was preceded by cognitive preparation or reduced self-focus. For broader context on how common speaking fears are, see our roundup on public speaking anxiety among students.
| Metric | Value | Source |
|---|---|---|
| Correlation between social anxiety (LSAS fear subscale) and own-voice liking, Arabic / English | r = -.228 / r = -.181 (both p < .05) | Jogia et al., International Journal of Psychology 2024 |
| Correlation between social anxiety and own-voice recognition, Arabic / English | r = -.174 (p < .05) / r = -.079 (not significant) | Jogia et al., International Journal of Psychology 2024 |
| Mean own-voice liking on a 1-10 scale, Arabic / English (176 Emirati women, mean age 20.69) | 5.54 / 5.40 | Jogia et al., International Journal of Psychology 2024 |
| Undergraduates who rated their own recorded voice against independent speech therapy student raters | 17; higher social anxiety linked to underestimating one’s voice qualities | Lundh et al., Cognitive Behaviour Therapy 2002 |
| Socially anxious participants in a single-session audio feedback experiment | 40; feedback with cognitive preparation produced less negative ratings | Nilsson et al., Journal of Behavior Therapy and Experimental Psychiatry 2011 |
| Participants in a follow-up randomized experiment on reduced self-focus | 41; reduced self-focus improved voice evaluations more than audio feedback only | Nilsson and Lundh, Cognitive Behaviour Therapy 2016 |
| Year of the Holzman, Berger and Rousey bilingual voice confrontation study | 1967 (Journal of Personality and Social Psychology 7(4)) | Holzman et al., Journal of Personality and Social Psychology 1967 |
Context note: all of these are correlational or small analogue studies, so they show association, not cause. The Lundh et al. 2002 study supported a cognitive distortion explanation (anxious people hear their voice as worse than it is) over a skill deficit explanation. The 2011 experiment and its 2016 follow-up both list the small sample as a limitation. Most recent available data for the Lund experiments: 2016.
6. Hearing Aids and the Altered Own Voice
For hearing aid users, voice confrontation is not an occasional event but a daily one. An earpiece that blocks the ear canal creates the occlusion effect, a boomy boost of low frequencies in your own voice, and hearing aid processing adds its own coloring. 40% of first-time users and 35.3% of experienced users rated their live voice as less pleasant with hearing aids in a Swedish study at Linköping University Hospital (Hengen et al., Trends in Hearing 2025). That is roughly double the share who reported moderate to severe own-voice problems, which means many users notice the change without calling it a problem.
The satisfaction data explain why own-voice quality matters commercially as well as clinically. If you are tracking why buyers give devices back, our roundup on OTC hearing aid return rates covers the adoption side.
| Metric | Value | Source |
|---|---|---|
| First-time hearing aid users with moderate to severe own-voice problems (n = 85) | 17.6% | Hengen et al., Trends in Hearing 2025 |
| Experienced hearing aid users with moderate to severe own-voice problems (n = 85) | 18.8% | Hengen et al., Trends in Hearing 2025 |
| First-time users who rated their live own voice as less pleasant with hearing aids | 40% | Hengen et al., Trends in Hearing 2025 |
| Experienced users who rated their live own voice as less pleasant with hearing aids | 35.3% | Hengen et al., Trends in Hearing 2025 |
| Hearing aid users satisfied with the sound of their own voice (survey of 384, mostly dome inserts) | 41% | Høydal 2017, as cited by Hengen et al., Trends in Hearing 2025 |
| Range of own-voice satisfaction across two hearing aid surveys | 41% to 73% | Høydal 2017 and Kochkin 2010, as cited by Hengen et al., Trends in Hearing 2025 |
| Preferred processing delay among 24 open-fit users (0.5 ms vs 5 ms vs 10 ms) | 0.5 ms | Súsonnudóttir et al., Trends in Hearing 2026 |
Context note: the Hengen et al. sample was older (mean age 70 for first-time users, 75 for experienced users and 70 for the 70 controls), and the questionnaire was written for the study and not yet validated. In Súsonnudóttir et al., the 0.5 ms delay also earned better ratings than 10 ms on tonality (attributes such as ‘metallic’ and ‘sharp’) and on own-voice quality.
7. Voice Clones and the Wish to Change Your Voice
Dislike of one’s recording does not stay passive: 50.6% of respondents without a history of voice disorder said they would consider an intervention to change their voice, with clarity and pitch the qualities they cared about most, in the same survey described in section 1. Software now offers a different kind of change, and the first experiments on cloned self-voices show that familiarity is the deciding factor in how a clone is judged.
In Rosi, Soopramanien and McGettigan, 25 pairs of close friends, relatives or partners rated recordings and clones of their own and each other’s voices. Familiar listeners spotted the clones easily and rated them less trustworthy than recordings, while strangers could not tell the difference. Separately, Jaggy, Schwan and Meyerhoff found that voices measured by an AI speaker-verification system as more similar to the listener’s own voice were rated as more likeable. Anyone building with voice cloning software gets a clear message from both: the person who knows the voice best is the hardest audience.
| Metric | Value | Source |
|---|---|---|
| Respondents without a voice disorder who would consider an intervention to change their voice | 50.6% | Naunheim et al., Laryngoscope 2023 |
| Clone detection sensitivity (d-prime), familiar listeners: own voice / friend’s voice | 2.13 (95% CI [1.84, 2.41]) / 1.91 (95% CI [1.60, 2.22]) | Rosi et al., Computers in Human Behavior: Artificial Humans 2025 |
| Clone detection sensitivity (d-prime), unfamiliar listeners | 0.14 (not different from chance, p = .420) | Rosi et al., Computers in Human Behavior: Artificial Humans 2025 |
| Trustworthiness of recordings vs clones, familiar listeners | Recordings rated higher, χ²(1) = 138.6, p < .001 | Rosi et al., Computers in Human Behavior: Artificial Humans 2025 |
| Effect of similarity to the listener’s own voice on likeability (93 German participants) | β = 0.11, 95% CI [0.05, 0.18], p < .001 | Jaggy et al., PLoS One 2025 |
| Effect of similarity to the listener’s own voice on trustworthiness (quadratic term) | β = 0.31, 95% CI [0.17, 0.45], p < .001 | Jaggy et al., PLoS One 2025 |
| Variance in likeability ratings explained by the own-voice similarity model | 19% | Jaggy et al., PLoS One 2025 |
Context note: d-prime is a sensitivity score in which 0 means guessing and higher values mean better detection. In the Rosi et al. study the familiar group was 47 young adults (mean age 22) recruited through University College London, so the result describes close friends listening carefully, not the general public.
Summary: Own-Voice Recording Dislike by the Numbers
| Metric | Value | Source |
|---|---|---|
| US adults dissatisfied with their own voice: listening to a recording vs normal conversation | 57.5% vs 38.8% | Naunheim et al., Laryngoscope 2023 |
| US adults very or somewhat uncomfortable hearing a recording of their voice | 43% | YouGov Daily Question, December 18, 2019 |
| US adults very comfortable hearing a recording of their voice | 23% | YouGov Daily Question, December 18, 2019 |
| Frequency range where bone conduction dominates one’s own voice | 1 to 2 kHz | Reinfeldt et al., JASA 2010 |
| Participants who chose the plain recording as most like their own voice | 2 of 10 | Kimura and Yotsumoto, PLoS One 2018 |
| Participants who chose the raw recording as most like their own voice (soundproof room) | 6 of 17 | Hosaka et al., Scientific Reports 2021 |
| Gap between perceived and produced own-voice pitch | 0.9 semitones | Neamaalkassis et al., Royal Society Open Science 2025 |
| Participants who spontaneously recognized their own voice in a rating task | 59 of 139 | Rosi et al., Cognition 2026 |
| Participants who rated their own voice blind in the classic self-enhancement study | 80 | Hughes and Harrison, Perception 2013 |
| Pitch shift that made one’s own voice less attractive than the plain recording | +/-20 Hz | Peng et al., Australian Journal of Psychology 2019 |
| Own-voice recognition, first vs second language | 85.6% vs 74.4% | Jogia et al., International Journal of Psychology 2024 |
| Self-voice hit rate in a control recognition task | 0.79 | Orepic et al., Royal Society Open Science 2023 |
| First-time hearing aid users with moderate to severe own-voice problems | 17.6% | Hengen et al., Trends in Hearing 2025 |
| Hearing aid users satisfied with their own voice (survey of 384) | 41% | Høydal 2017, via Hengen et al. 2025 |
| Preferred hearing aid processing delay for own-voice perception (24 open-fit users) | 0.5 ms | Súsonnudóttir et al., Trends in Hearing 2026 |
| Clone detection d-prime for one’s own voice, familiar vs unfamiliar listeners | 2.13 vs 0.14 | Rosi et al., Computers in Human Behavior: Artificial Humans 2025 |
| Respondents without a voice disorder open to changing their voice | 50.6% | Naunheim et al., Laryngoscope 2023 |
Methodology and Sources
- Naunheim, Puka and Huston, ‘Do You Like Your Voice? A Population-Based Survey of Voice Satisfaction and Voice Enhancement’, Laryngoscope 133(12), 2023, with the ENT Today summary used for the sample’s age and sex split.
- YouGov Daily Question, ‘How comfortable or uncomfortable are you listening to a recording of your own voice?’, December 18, 2019.
- Jogia, Thomas, Barbato and Bentall, ‘Social anxiety, voice confrontation and voice recognition: A bilingual exploration’, International Journal of Psychology 59(6), 2024, also the source of the Chong et al. (Journal of Voice 2024) figures.
- Reinfeldt, Östli, Håkansson and Stenfelt, ‘Hearing one’s own voice during phoneme vocalization’, JASA 128(2), 2010.
- Methods to experimentally characterize the own-voice-generated objective occlusion effect induced by hearables, Acta Acustica 9, article 73, 2025.
- Kimura and Yotsumoto, ‘Auditory traits of own voice’, PLoS One 2018.
- Hosaka, Kimura and Yotsumoto, ‘Neural representations of own-voice in the human auditory cortex’, Scientific Reports 11:591, 2021.
- Neamaalkassis et al., ‘The fundamental frequencies of our own voice’, Royal Society Open Science 12(2), 2025.
- Suzuki, Kuroda and Nagai, ‘The Positively Reconstructed Self-Voice’, PsyArXiv preprint 2026.
- Hughes and Harrison, ‘I like my voice better: Self-enhancement bias in perceptions of voice attractiveness’, Perception 42(9), 2013.
- Peng et al., ‘One’s own and similar voices are more attractive than other voices’, Australian Journal of Psychology 71(3), 2019.
- Peng, Hu, Wang and Liu, self-enhancement effect of voice attractiveness evaluation, Scandinavian Journal of Psychology 61(5), 2020.
- Rosi, Lavan and McGettigan, ‘Self-recognition shapes evaluations of (self-) voice attractiveness’, Cognition 272, 2026.
- Rosi, Soopramanien and McGettigan, ‘Perception and social evaluation of cloned and recorded voices’, Computers in Human Behavior: Artificial Humans 4, 2025.
- Xu, Homae, Hashimoto and Hagiwara, ‘Acoustic cues for the recognition of self-voice and other-voice’, Frontiers in Psychology 4:735, 2013.
- Orepic, Kannape, Faivre and Blanke, ‘Bone conduction facilitates self-other voice discrimination’, Royal Society Open Science 10(2), 2023.
- Kirk and Cunningham, ‘Listen to yourself! Prioritization of self-associated and own voice cues’, British Journal of Psychology 116(1), 2025.
- Zheng, MacDonald, Munhall and Johnsrude, ‘Perceiving a stranger’s voice as being one’s own’, PLoS One 2011.
- Lundh et al., ‘Social Anxiety is Associated with a Negatively Distorted Perception of One’s Own Voice’, Cognitive Behaviour Therapy 31(1), 2002.
- Nilsson, Lundh, Faghihi and Roth-Andersson, audio feedback with cognitive preparation, Journal of Behavior Therapy and Experimental Psychiatry 42(4), 2011.
- Nilsson and Lundh, ‘Audio Feedback with Reduced Self-focus as an Intervention for Social Anxiety’, Cognitive Behaviour Therapy 45(2), 2016.
- Holzman, Berger and Rousey, ‘Voice confrontation: a bilingual study’, Journal of Personality and Social Psychology 7(4), 1967.
- Hengen, Lundeborg Hammarström and Stenfelt, ‘Effect of Hearing Aids on Phonation and Perceived Voice Qualities’, Trends in Hearing 29, 2025, also the source of the Høydal 2017 and Kochkin 2010 satisfaction figures.
- Súsonnudóttir et al., ‘Own-Voice Perception with Different Processing Delays in Open-Fit Hearing Aids’, Trends in Hearing 30, 2026.
- Jaggy, Schwan and Meyerhoff, ‘AI-determined similarity increases likability and trustworthiness of human voices’, PLoS One 20(3), 2025.
- Data watch: Peng et al. 2020 report self-enhancement whether or not people recognized their voice, while Rosi et al. 2026 find that non-recognizers rated their own voice as less attractive; treat the ‘we secretly like our voice’ claim as contested. Some press coverage of the Laryngoscope survey printed its headline figures as ‘8%’ and ‘5%’, a rendering error; the abstract gives 38.8% and 57.5%. The Chong et al. shares (31%, 24%, 44%) are taken from Jogia et al. and sum to 99%, likely from rounding. Jogia et al. report the Arabic vs English recognition test (χ²(1, N = 98) = 5.1848, p = 0.022) with N = 98 while the full sample was 176, which the paper does not explain, so the 85.6% and 74.4% recognition rates may rest on a subsample. The Rosi et al. 2025 paper prints a confidence interval for unfamiliar listeners that does not bracket its mean, likely a typo, so we quote only the mean and p value. The Acta Acustica 2025 study warns that the common simultaneous method of measuring the occlusion effect carries systematic errors of up to about 3-4 dB. The Kimura and Yotsumoto Experiment 2 counts are ambiguous in the paper, so only Experiment 1 is used. The YouGov figures date from 2019, and Suzuki et al. 2026 is a preprint that has not completed peer review.
Last updated: November 8, 2026. We update this roundup quarterly, and the next refresh is expected when the peer-reviewed version of the Suzuki et al. reconstructed self-voice study is published.