You send a voice note, play it back on speaker by accident, and for a second you have no idea who is talking. It is your voice. It is just not the one you know.
Your recorded voice sounds thinner and higher for one reason: the recording is missing bone conduction (sound travelling from your vocal folds to your inner ear straight through the bones of your skull instead of through the air). Bone adds low frequencies, so the voice you hear from the inside feels deeper and fuller. A microphone never picks that part up. It only records what travels through the air. So the recording is not the odd one out. The odd one out is the version only you have ever heard.
This is not a niche problem. On one messaging app alone people send around 7 billion voice messages a day (a figure WhatsApp published in 2022). We talk into microphones constantly, and most of us wince at the result. The strangest finding on this whole subject came from an experiment where people were not told whose voice they were rating. We will get back to it.
Why does your voice sound higher in a recording?
Because you hear yourself through two paths at once, and the microphone only gets one of them.
The first path is air: sound leaves your mouth, travels around your head and reaches your ear the normal way. That is exactly what a recorder captures. The second path is shorter. Your vocal folds vibrate, and that vibration runs through skull bone and soft tissue straight into your inner ear, skipping the air entirely. That second channel is bone conduction.
Researchers at Linköping University in Sweden measured both paths in 38 people while they produced different speech sounds; the work appeared in the Journal of the Acoustical Society of America in 2010. The bone path contributed most in the low and middle frequencies, while air took over higher up. Hence the effect: the inside voice is bassier, the recorded one is thinner and brighter.
You can test it in five seconds. Cover your ears with your palms and say a sentence out loud. Your voice thickens instantly, because you muffled the air path and left the bone path almost on its own. Closed headphones do roughly the same thing, which is why your own voice sounds oddly loud inside them.
Here is a simple way to picture it. The voice in your head is a track playing through speakers with the subwoofer on. The recording is the same track with the subwoofer off. Same notes, different feeling.
Which voice is the real one, the recording or the one in your head?
Both are real, but other people only ever get the recorded one.
The bone-conducted part exists strictly inside your own skull. Your colleague on a call, the cashier at the shop, the friend who got your voice note: none of them has ever heard it. To them your voice has always sounded like the recording, and it does not bother them at all.
Neuroscientists at EPFL in Lausanne, Switzerland tested how deeply that extra layer is wired into recognising yourself. In a study published in Royal Society Open Science in 2023, participants judged voice morphs (blends of their own voice with someone else's, mixed in ratios from 15 to 85 percent) and had to say whose voice it was. Some sounds came through headphones, others through a vibrating device pressed against the skull. Bone delivery improved recognition of their own voice specifically, and made no difference when they judged other familiar voices.
What that means for you: your brain treats the bone-conducted layer as a tag that says "this is me". Recordings have no tag, so the first reaction is honest: somebody else is talking.
The limits are worth stating plainly. The samples were small: three experiments with 16, 16 and 52 people. The setting was a lab. The authors note themselves that they could not fully separate the acoustic effect from the feeling of vibration on the skin.
There is a second layer to this, and it is psychological. A team at the Max Planck Institute for Empirical Aesthetics in Frankfurt separated two things in 2025: the objective pitch of your voice (the fundamental frequency, meaning the lowest rate at which your vocal folds vibrate, which any analyser can measure) and the subjective pitch you think of as yours. In their experiment people were worse at noticing small pitch changes near either of those than near a control pitch. In plain terms, your brain keeps an internal portrait of your voice and stops nitpicking details inside it.
So much for the physics. But if a recording is just you with the subwoofer off, why does it make you flinch?
Why is hearing your own voice so uncomfortable?
Because a recording is the first time you hear yourself from the outside, and you hear far more than you bargained for.
The phenomenon has a name: voice confrontation. Psychologists Philip Holzman and Clyde Rousey established the term back in 1966 while studying how people react to their own recordings. The discomfort is not really about tone. A recording exposes what you never monitor while speaking: the rush, the hesitation, the irritation, the pauses that felt short at the time.
Neel Bhatt, an ear, nose and throat specialist at the University of Washington, describes the reaction as a clash between self-image and reality, and points to a telling result: in a 2005 study, patients rated the quality of their own recorded speech far more harshly than clinicians rated the very same recordings. The gap was not in the sound. It was in who was listening.
In real life it looks like this. You record a 40-second voice note for a colleague, play it back before sending, and count six filler words, two false starts and a laugh that came out weirdly high. Your colleague would have heard a normal person explaining a task. You hear a disaster.
From here on, everything hangs on the "this is me" tag. Take it away, and the ratings flip.
Secretly, you like your own voice: the experiment that flips it
People rate their own voice higher than other voices, as long as they do not know they are listening to themselves.
Psychologists Susan Hughes and Marissa Harrison recruited 80 participants and asked them to rate how attractive a set of recorded voices sounded. Nobody was told that their own voice had been slipped into the set. The result, published in Perception in 2013: people rated their own voice higher than other listeners rated it, and higher than they rated everyone else's. The authors called it vocal implicit egotism, a form of self-enhancement.
Put the picture together. As pure sound, other people tend to like your voice, and so do you, right up until recognition kicks in. The problem is not the frequencies and not the microphone. The problem is the label.
Which means the thing to work on is not your voice. It is the habit of hearing it.
How do you get used to your own recorded voice?
Listen to yourself often and in small doses, instead of once with a full post-mortem.
What does the work here is the mere exposure effect (a well-documented pattern where a neutral thing becomes more likeable the more often you encounter it). Your recorded voice is a new object to you. The first few times it registers as foreign, then your brain stops flagging it as news. It is the same mechanism that turns an annoying song into a perfectly fine one within a week.
The usual failure mode goes the other way. Someone records themselves once, recoils, deletes the file, and avoids microphones for the next six months. Avoidance gives the brain no chance to adapt, so every call with the camera on arrives with the same tension as the first one.
But the honest advice to "listen to yourself more" has a weak spot. A voice recorder is a boring gym. There is no task in it, no game, and no reason to open it again tomorrow. Most people quit on day two.
Give the voice a job, though, and the whole thing changes.
Why does dubbing a video beat a voice recorder?
Because you stop assessing yourself and start trying to land someone else's line.
Dubbing (re-voicing an existing video in your own voice) has been tested for years as a classroom technique rather than a party trick. The logic matches your discomfort exactly: you hear a model, you copy it, you hear the result.
What the research shows:
- Hong Kong, 2024. Min Xie and Alex Tse of the University of Hong Kong compared two approaches with fifth-graders: 104 children practised English through online video dubbing, 102 learned the same content the usual way. After 12 sessions the difference in speaking proficiency was statistically significant (F = 8.291, p = 0.004), meaning chance is an unlikely explanation.
- Vietnam, 2025. Van Tuyet Thi Phan and colleagues ran 10 weeks of dubbing with 30 eleven-year-olds. Overall pronunciation scores rose from 14.86 to 16.50, with intonation improving most (3.36 to 4.30). Almost all of them (96.67 percent) said it helped, and 83.33 percent said it made practice more motivating.
- A caution on the Vietnamese study. The authors list the limits themselves: a small group, no control group, a short run. One study like that proves nothing on its own, but it points the same way as the Hong Kong experiment, which did have a control group.
One honest caveat, so nothing gets oversold. Both studies measured speaking in a foreign language, not how self-conscious anyone felt. There is no direct evidence yet that dubbing cures a dislike of your own voice. What does hold up separately is the mechanism: repeated contact strips a recording of its novelty, and having a task moves your attention off self-assessment and onto timing.
How to dub a video in your own voice in 5 minutes
The easiest way in is a short clip and a handful of lines.
We built Awe Voice for exactly that: a free dubbing game that runs in your browser. It works like this:
- Pick a video. Your own clip from your phone, laptop or a cloud drive.
- Mark the lines. Up to five, fifteen seconds each. Listen to the original, say it in your own voice, re-record as often as you like.
- Watch it, then keep it. You hear your own voice only at the very end, running over the whole clip, and you download the file.
That last part is the point. You cannot hear yourself while recording. During a take your attention is on someone else's line and on hitting the timing, not on picking apart your own tone. The meeting with the recording happens once, at the end, and it arrives attached to something you made.
On privacy, because the microphone question always comes first: neither the video nor your recording is uploaded. Everything lives in the open tab and disappears when you close it. The only person who can share the result is you, after downloading the file. That is not developer paranoia either: a few seconds of recorded speech is now enough to synthesise a voice, which we covered in our piece on how many seconds it takes to clone a voice.
Five mistakes people make on their first dub
Almost all of them are technical, not a matter of talent.
- Recording without headphones. The mic will pick up the original along with you, and the two tracks pile on top of each other in the final cut.
- Going quiet out of embarrassment. Watch the level bar: it should reach at least the middle. A timid take sounds unsure even where you performed with confidence.
- Choosing long lines. A 15-second phrase is hard to hold in your head. Five to seven seconds lands on the first or second try.
- Pressing your mouth against the mic. You get rumble and popping on hard consonants. A palm's width between mouth and phone is about right.
- Chasing perfection on line one. The funny part comes from the gap between what you thought you did and what actually came out. A flawless dub is usually a boring one.
What to do today
Open any clip of yours that runs about a minute, mark a single ten-second line, and voice it. That takes under five minutes and gives you the one thing a voice recorder cannot: you hear yourself as part of a scene rather than as an object under review.
And that voice note that made you flinch at the start stops sounding like a stranger. It never was one. You had simply never heard yourself with the subwoofer off.
Sources
- Reinfeldt S. et al. Hearing one's own voice during phoneme vocalization. Journal of the Acoustical Society of America (2010)
- Orepic P., Kannape O., Faivre N., Blanke O. Bone conduction facilitates self-other voice discrimination. Royal Society Open Science (2023)
- Neamaalkassis H. et al. The fundamental frequencies of our own voice. Royal Society Open Science (2025)
- Hughes S., Harrison M. I like my voice better: self-enhancement bias in perceptions of voice attractiveness. Perception (2013)
- Bhatt N. Why do we hate the sound of our own voices? The Conversation, University of Washington (2021)
- Xie M., Tse A. Impact of online video dubbing activities on grade 5 students' pronunciation, accuracy and fluency. ICCE, University of Hong Kong (2024)
- Phan V. T. T., Nguyen T. Q., Nguyen K. D. The effects of video dubbing on EFL learners' pronunciation. TESL-EJ (2025)
- TechCrunch, citing WhatsApp: 7 billion voice messages sent daily (2022)
