Before you were born
The first voice you ever heard
For months before you were born, you could hear. Mostly, you heard one person. Put on headphones to hear a mother's voice the way it reaches the child inside her, rebuilt from studies that placed microphones in the womb. Then record your own.
The sound starts as you scroll, at its real loudness: low and quiet, as it was for you. You can boost it at any time. Phone speakers lose most of it.
Before there was anything to hear
Week 12. Around you there is already sound: her blood, her breathing, her voice. But there is no evidence yet that you can hear it.
No study has seen a fetus react to sound this early, so the page stays silent here.
The first sound was one note
Week 19
In 1994, the psychologists Hepper and Shahidullah played tones through a loudspeaker placed on pregnant women's bellies and watched the fetuses on ultrasound. The first to move did so at 19 weeks, and only for one tone: 500 hertz, close to the B above middle C.
What you hear now is her voice narrowed to that one band. This illustrates the experiment: 500 hertz was the first tone that got a response, which does not mean the fetus heard nothing else.
Week 27
The range grows downward first. At this age 96% of the fetuses in the study moved for the low tones of 250 and 500 hertz. None moved for 1,000 hertz.
Week 33
By now every fetus answered 1,000 hertz, and by week 35, 3,000. The ear also grows more sensitive: between week 19 and week 35, the tones needed to get a response became 20 to 30 decibels quieter.
This is close to the ear you were born with, listening from inside her.
What it sounds like in there
You have probably imagined it like this: a heartbeat, steady and close. Soft toys for newborns are sold with this sound built in.
In 1990, a team in Ontario placed a hydrophone beside the necks of babies about to be born, in ten women in labour. In eight of them, there was no heart sound to hear at all.
What they recorded was a low rumble, most of it below 100 hertz; now and then a burst from her intestines; and, well above the background, her voice.
A fetus may still sense her heartbeat: in 2006, an Italian team recorded fetal brain responses to it in 8 of 12 cases. But the steady thump filling the womb is not what the microphones found.
Studies in pregnant sheep, whose lambs also start hearing before birth, found the same: voices from outside came through clearly, not drowned out by the mother's heart.
Hers was the voice that got louder
In 1992, a team in Florida measured sound inside the uterus of eight women in labour. Voices from the room lost 2 to 3 decibels on the way in. Her own voice gained 5.2.
It was the only voice that did not have to come in from outside.
Listen from three places
The voice is a woman reading Louisa May Alcott's poem Lullaby, recorded for LibriVox.
Choose a voice and a place. This is a reconstruction: the water curve comes from women in labour, the inner-ear curve from fetal sheep. The sound follows those curves closely; the curves themselves are the uncertain part.
The surprise is where the muffling happens. In the water around you, the loss is small: 10 decibels at most, at 4,000 hertz, and the lowest notes are slightly louder than outside.
The large loss comes between the water and the inner ear. In fetal sheep, sound reached the inner ear mainly through the bones of the skull, and lost 11 decibels at 125 hertz but 45 at 2,000.
What survived was the melody. Listeners understood almost 99% of sentences recorded in a ewe's uterus, and 41% of the same sentences recorded at the lamb's inner ear. In humans, Denis Querleu's team estimated that only about 30% of the speech-sound detail even reaches the fluid around the fetus, while intonation arrives almost intact.
You heard how she said things more than what she said.
Now, your voice
Say something to someone who has not been born yet
Record up to 12 seconds: a name, a sentence, a line of a song. Then hear it the way the measurements suggest a baby inside would.
The recording stays on your device and disappears when you close the page.
Recording needs a microphone, and this device does not offer one here. The same filters shape her voice in the panel above.
You were already learning her
In Finland, a group of mothers played the made-up word tatata from week 29 until birth: 25,740 times on average. After birth, the babies' brains responded more strongly than other babies' when its middle syllable changed pitch.
In another group, mothers played Twinkle, Twinkle, Little Star five times a week in the last three months. Four months after birth, their brains still responded to it more strongly than the brains of babies who had not heard it. This is the tune.
In 2003, sixty fetuses near term heard a recording through a loudspeaker held above the belly. Some heard their mother; others, a woman they had never heard. Heart rates rose for the mother and fell for the stranger. The voice played for two minutes; the change lasted four, well after it stopped.
Then the water is gone
Week 40. You have been listening to her, a little more clearly each week, since week 19.
And the first thing that changes is her voice. For the first time, the consonants come into focus.
In 1980, Anthony DeCasper and William Fifer let newborns choose, by the way they sucked on a teat, between a recording of their mother's voice and another woman's. They learned the trick, and they worked for their mother's voice.
Your first cry may have had her accent
In 2009, Birgit Mampe and colleagues in Germany recorded the cries of 30 French and 30 German newborns, 2 to 5 days old. The French babies' cries mostly rose in pitch toward the end. The German babies' mostly fell. Each, they concluded, followed the typical melody of the language their mothers spoke. Later studies found similar patterns in other languages; others have questioned how strong the effect is.
In Sweden and the United States, babies 33 hours old on average already treated the vowels of their mother's language as familiar.
Under everything you have heard since, there is a first voice. It reached you mostly as a melody, and you learned it well enough to cry in it.
If you can still call her, you know who to call. And if someone is listening to you from the inside right now, keep talking. They are learning your voice.
Sources and method
Every sound on this page passes through filters set to published measurements. In the water: the human hydrophone study of Richards and colleagues, which gives +3.7 dB at 125 Hz and −10 dB at 4,000 Hz; the bands in between are drawn on a straight line. At the inner ear: the fetal sheep measurements of Gerhardt and colleagues, 125 to 2,000 Hz, extended above 2,000 Hz at the last measured slope (-6.8 dB per octave). No one has measured this step in a human fetus; sheep are the standard model because their hearing develops before birth, as ours does.
Boost sound plays the inner-ear curve 11.1 dB louder than measured, and voices a further 10.5 dB; the shape is unchanged. These boosts apply only when you press Boost sound. By default, voices at the inner ear play about 16 dB below the room, as measured. Her voice gets the same curve as sound from outside, 8.4 dB stronger, because no one has measured separately how a mother's own voice reaches the fetal inner ear.
The weeks follow the tones fetuses first moved to in Hepper and Shahidullah's study (500 Hz at 19 weeks; 250 and 500 Hz but not 1,000 at 27; 1,000 Hz by 33; 3,000 Hz by 35). The weeks in between are drawn smoothly. The ear's growing sensitivity is shown over 14 dB rather than the 20 to 30 dB measured, so that the earliest weeks stay audible. The background is built from what Benzaquen and colleagues recorded: most of the energy below 100 Hz, occasional bursts from the intestines, and no heartbeat. The hummed tune is synthesised. Most sources were read in their published abstracts; Partanen and colleagues (PNAS) was read in full.
- Hepper & Shahidullah (1994), Development of fetal hearing, Arch Dis Child Fetal Neonatal Ed 71:F81–87 doi.org/10.1136/fn.71.2.f81
- Richards, Frentzen, Gerhardt, McCann & Abrams (1992), Sound levels in the human uterus, Obstet Gynecol 80:186–190 pubmed.ncbi.nlm.nih.gov/1635729/
- Benzaquen, Gagnon, Hunse & Foreman (1990), The intrauterine sound environment of the human fetus during labor, Am J Obstet Gynecol 163:484–490 doi.org/10.1016/0002-9378(90)91180-k
- Gerhardt, Otto, Abrams et al. (1992), Cochlear microphonics recorded from fetal and newborn sheep, Am J Otolaryngol 13:226–233 doi.org/10.1016/0196-0709(92)90026-p
- Gerhardt, Huang, Arrington et al. (1996), Fetal sheep in utero hear through bone conduction, Am J Otolaryngol 17:374–379 doi.org/10.1016/s0196-0709(96)90069-1
- Gerhardt, Abrams & Oliver (1990), Sound environment of the fetal sheep, Am J Obstet Gynecol 162:282–287 doi.org/10.1016/0002-9378(90)90866-6
- Gerhardt & Abrams (2000), Fetal exposures to sound and vibroacoustic stimulation, J Perinatol 20:S21–30 doi.org/10.1038/sj.jp.7200446
- Smith, Gerhardt, Griffiths, Huang & Abrams (2003), Intelligibility of sentences recorded from the uterus of a pregnant ewe and from the fetal inner ear, Audiol Neurootol 8:347–353 doi.org/10.1159/000073519
- Querleu, Renard, Versyp, Paris-Delrue & Crèpin (1988), Fetal hearing, Eur J Obstet Gynecol Reprod Biol 28:191–212 doi.org/10.1016/0028-2243(88)90030-5
- Armitage, Baldwin & Vince (1980), The fetal sound environment of sheep, Science 208:1173–1174 doi.org/10.1126/science.7375927
- Kisilevsky, Hains, Lee et al. (2003), Effects of experience on fetal voice recognition, Psychol Sci 14:220–224 doi.org/10.1111/1467-9280.02435
- Partanen, Kujala, Näätänen et al. (2013), Learning-induced neural plasticity of speech processing before birth, PNAS 110:15145–15150 doi.org/10.1073/pnas.1302159110
- Partanen, Kujala, Tervaniemi & Huotilainen (2013), Prenatal music exposure induces long-term neural effects, PLoS One 8:e78946 doi.org/10.1371/journal.pone.0078946
- DeCasper & Fifer (1980), Of human bonding: newborns prefer their mothers' voices, Science 208:1174–1176 doi.org/10.1126/science.7375928
- Moon, Lagercrantz & Kuhl (2013), Language experienced in utero affects vowel perception after birth, Acta Paediatr 102:156–160 doi.org/10.1111/apa.12098
- Mampe, Friederici, Christophe & Wermke (2009), Newborns' cry melody is shaped by their native language, Curr Biol 19:1994–1997 doi.org/10.1016/j.cub.2009.09.064
- Porcaro, Zappasodi, Barbati et al. (2006), Fetal auditory responses to external sounds and mother's heart beat, Brain Res 1101:51–58 doi.org/10.1016/j.brainres.2006.04.134
- Hervais-Adelman & Townsend (2025), How did vocal communication come to dominate human language? A view from the womb, PLoS Biol 23:e3003141 doi.org/10.1371/journal.pbio.3003141
- The voice: Lullaby by Louisa May Alcott (1832–1888), read by Heather Dalton, LibriVox recording librivox.org/lullaby-by-louisa-may-alcott/