Do binaural beats work? What the studies found

As a sound, yes: play a slightly different tone in each ear and your brain hears a third tone, pulsing at the difference. Whether that changes how you think or feel is far less settled. A 2023 review found more brain-recording studies against the idea that brain rhythms follow the beat than for it, and pooled effects on memory and attention rest on small, mixed studies.

Ready to play Alpha Binaural Beat 10 Hz

Open the listening page

Headphones for binaural. Keep the volume low and stop if anything feels uncomfortable; take care if you have tinnitus or sensitive hearing. Never while driving. Harmonance is not a medical device.

Eleven gold ribbons in dark space swelling into one tall wave at the centre and settling into gentle ripples at both edges — Eleven ribbons in the alpha band's gold, carrying two close waves at once, so the bundle swells at the centre and settles at the edges the way a beat rises and falls.

What are binaural beats?

Play 108 Hz in your left ear and 118 Hz in your right, and you hear something neither ear is given: a soft pulse, ten times a second. That pulse is a binaural beat. The two tones never mix in the air, so there is no 10 Hz sound for a microphone to pick up. The beat is made inside your head. One research team, recording the brain's response to it, called it "a signal that is not present in the stimulus, but generated within the brain" (Schwarz and Taylor).

It forms early, in the brainstem, where the signals from your two ears first meet: a review places it most likely in the superior olivary nuclei (Chaieb and colleagues). H. W. Dove first described the effect in 1839 (Ingendoh and colleagues), and it reached a wide readership in 1973, when Gerald Oster wrote it up in Scientific American as a sensation that "may show how certain sounds are processed by the brain".

It needs low tones. Oster heard the beats only when the carrier tone was below 1,000 Hz, as Chaieb and colleagues report. In 1950 J. C. R. Licklider and his colleagues found the widest range of beat rates near 400 Hz, where two tones about 35 Hz apart still beat (Licklider, Webster and Hedlun). In 1969 Perrott and Nelson found beats easiest to detect at 500 Hz, and no reliable detections above 1,500 Hz (Perrott and Nelson). Every recording in our Binaural Beats collection keeps 108 Hz in the left ear, well inside that range.

Two close tones played through one speaker are a different thing. They mix in the air and really do add and cancel, so the loudness swells and fades as sound: a monaural, or acoustic, beat. A binaural beat needs headphones, one tone to each ear. Isochronic tones are a third kind, a single tone switched on and off (isochronic tones and binaural beats compared).

What does a binaural beat sound like?

It depends on the gap between the two tones. Below about 3 Hz the sound seems to rotate in your head from ear to ear. Up to about 20 Hz you hear a tone that rises and falls in loudness, the familiar pulse. Beyond 20 Hz it turns rough, and in all the beat is heard from roughly 1 to 30 Hz (Ingendoh and colleagues). Listening studies agree on the order: a sense of motion, then loudness fluctuation, then roughness, until the two tones are simply heard apart, one in each ear (Grose and colleagues).

It is a quiet effect. All twelve adults in that study heard the beats, but the authors judged their perceptual strength "relatively weak", and they note that the beat can fade after several minutes of continuous listening. In another study of 15 adults, people noticed a binaural beat less reliably than an acoustic one, even at slow rates: detection fell away above 15 Hz, was one in two at 35 Hz, and at 40 Hz was "close to chance level". The acoustic beat was heard almost perfectly up to 60 Hz (Ross and colleagues).

A beat rate scale from 1 to 80 Hz: moving below 3 Hz, pulsing to 20, rough to 30, then hard to hear, over the five bands and our 13 recordings — What a binaural beat sounds like at each rate, with the brainwave bands and our 13 recordings beneath: squares for the steady beats, arrows for the sessions that sweep. The scale is logarithmic, so each doubling takes the same width, and the edges between sounds are approximate.

The odd one out is 40 Hz. Brain recordings respond to a binaural beat most strongly at about that rate (Ross and colleagues), yet it is a rate the same listeners could barely tell was beating. Our 40 Hz gamma beat and the rising gamma session sit there, so listen for a faint roughness rather than a clear pulse.

Do binaural beats change your brainwaves?

This is the idea behind most claims made for binaural beats: that the brain's own rhythms fall into step with the beat, so a 10 Hz beat would draw it towards 10 Hz alpha. Researchers call it brainwave entrainment, and the evidence for it is split. A 2023 systematic review found 14 EEG studies: five reported results in line with entrainment, eight contradicted it and one was mixed. The studies differed so much in how they played the beats and measured the brain that, the authors concluded, the question cannot be settled yet (Ingendoh and colleagues).

The brain does respond to the beat. Recordings pick up a response at the beat rate, largest near 40 Hz (Ross and colleagues). A response in the hearing parts of the brain is not the whole brain changing state, though. A blinded study that compared binaural beats with monaural beats, which are known to drive the cortex, found that binaural beats "weakly entrain the cortex" by comparison, and found no change in mood (Orozco Perez and colleagues). A high-density EEG study of 31 people found no change in brain rhythms or vigilance during two minutes of 7 or 16 Hz beats, compared with white noise (Goodin and colleagues). The guide to brain waves, band by band covers what each band is.

Do binaural beats work for focus and memory?

Pooled together, the studies look better than most single trials. A 2019 meta-analysis combined 22 studies and 35 effect sizes and reported an average effect of 0.45 in standard units across memory, attention and other measures, which its authors call medium (Garcia-Argibay and colleagues). It also found that masking the beat with white or pink noise made no difference, that listening before a task did better than listening only during it, and that longer listening went with larger effects. A 2023 meta-analysis of 15 studies of memory and attention reached a similar 0.40, while noting conflicting results on memory and attention tasks, especially for theta and beta beats (Basu and Banerjee).

On Jacob Cohen's yardstick for effects like these, 0.2 is small, 0.5 medium and 0.8 large, so both averages fall between small and medium (Sullivan and Feinn).

Fourteen squares, five gold, one half gold and eight empty, above four pooled effects from 0.40 to 1.38, the widest ranging from 0.03 to 1.99 — What the reviews found. Top: the 14 EEG studies in a 2023 review, by whether brain rhythms followed the beat. Bottom: four pooled effects in standard units, against Cohen's marks for small, medium and large, with their 95 per cent range where the review gives one; the 2019 and 2023 summaries give none.

The numbers behind the chart
Review	What it pooled	Result
Ingendoh and colleagues, 2023	14 EEG studies: do brain rhythms follow the beat?	5 yes, 8 no, 1 mixed
Garcia-Argibay and colleagues, 2019	22 studies of memory, attention and more	0.45
Basu and Banerjee, 2023	15 studies of memory and attention	0.40
Xiong and colleagues, 2025	Nerves around an operation, against blank audio: 14 trials, 1,047 people	1.38 (0.87 to 1.89)
Xiong and colleagues, 2025	Nerves around an operation, against other audio: 8 trials, 598 people	1.01 (0.03 to 1.99)

Single trials pull both ways. Among 58 adults who took an attention test at home, 40 Hz beats made no difference against a plain control tone (Leistiko and colleagues). Among 25 students, 40 Hz beats went with fewer errors on an attention task than pink noise, but their EEG could not confirm that brain rhythms followed the beat (Engelbregt and colleagues). And a 2017 study of 40 Hz beats was retracted after a university investigation found evidence of manipulated data: 88 people were tested and only 40 analysed, with no adequate explanation, and one of its three authors disagrees with the retraction (the retraction note).

If you would like a beat in the background while you work, the need to focus door gathers the tones people reach for, and the collection has rising beta and gamma sessions. Use them as a backdrop, not a method.

Do binaural beats work for sleep?

Sleep is the most common reason people use them. In the Global Drug Survey of 2021, nearly three in four people who had used binaural beats said they did so to relax or fall asleep (Barratt and colleagues). The evidence is thin. The largest test gave 221 people a phone app and three tries each at falling asleep to the same steady sound: with binaural beats, with monaural beats, or with no beats at all. How long they took to fall asleep did not differ between the three (Shumov and colleagues).

A smaller sleep-lab study points the other way. In 24 people, a 3 Hz beat played during light sleep went with more time in N3, the deepest stage of sleep, than a silent sham did (Jirakittayakorn and Wongsawat). One small study cannot settle it. If you would like to listen at night, the falling delta session eases from 3.5 to 1.5 Hz over nine hours, and the can't sleep door gathers other gentle tones. Keep the volume low.

What do the hospital studies show?

The steadiest results come from people around the time of an operation, asked how anxious they felt. A 2025 meta-analysis pooled 15 trials. Against blank audio, 14 trials with 1,047 people found a large average difference, 1.38 in standard units. Against other audio, 8 trials with 598 people found a smaller one, 1.01, with a range so wide, from 0.03 to 1.99, that the true figure could be close to nothing. The trials disagreed sharply with each other, and the authors call for standard methods and more high-quality trials (Xiong and colleagues).

Those numbers describe how people rated their own nerves in hospital, often against silence. They cannot say what listening at home would do.

Are binaural beats safe?

A 2024 review found that "no study has reported any side effects during or after listening", advised keeping the level below 85 dB because loud sound can damage hearing over time, and recommended that anyone with epilepsy check with their doctor before trying binaural beats. It also notes that some people find the repetition uncomfortable (Mirmohamadi and colleagues). That covers studies at 40 Hz too, such as the attention trials above. The World Health Organization puts safe listening at 80 dB for up to 40 hours a week, and suggests keeping your device at no more than 60 per cent of its maximum volume (WHO, safe listening).

The darker side people search for is the idea of binaural beats as "digital drugs". In the Global Drug Survey of 2021, 5.3 per cent of 30,896 respondents had used binaural beats to try to reach altered states, and 11.7 per cent of those hoped for an effect like that of other drugs. Most found them on video sites, on their phones (Barratt and colleagues). The research above gives no sign of an effect of that size: the beat itself is, in one review's words, "a weak percept" (Chaieb and colleagues). Never listen while driving or doing anything that needs your full attention, and stop if anything feels uncomfortable.

How do you listen to binaural beats correctly?

Use headphones or earbuds, one tone to each ear. Through a speaker the two tones mix in the air and you hear an acoustic beat instead.
On AirPods, turn off Spatial Audio or Spatialize Stereo (Apple Support), and leave Mono Audio off: it plays "the same content in both ears", which removes the difference the beat needs (AirPods User Guide).
Keep the volume low, just loud enough to hear both tones.
Leave out masking noise: in the 2019 meta-analysis, beats with and without white or pink noise did about as well.
Sit or lie comfortably and notice how the sound moves or pulses. Some people hear it clearly, some faintly, and it can fade after a few minutes.

How long should you listen to binaural beats?

No study has tested the right length. Trials have played beats for as little as two minutes and for as long as people took to fall asleep, and the 2019 meta-analysis found larger effects with longer listening, and with listening before a task rather than only during it (Garcia-Argibay and colleagues). Our recordings run for nine hours so they can sit in the background; nobody needs to hear all nine. Keep within the WHO's weekly guide above.

Which binaural beat should you try?

Choose by what you would like to hear, not by a promised state. The diagram above maps our 13 recordings onto what the research says you will hear.

A slow, rocking movement: the delta beats at 1.618 Hz and 2 Hz, and the falling delta session, which spends three quarters of its nine hours below 3 Hz.
A clear, gentle pulse: the theta and alpha beats at 4 Hz, 6 Hz, 7.83 Hz and 10 Hz, and the theta and alpha sessions, all below 15 Hz, where the beat is easiest to notice.
A faster, rougher texture: the 20 Hz beta beat and the rising beta session, which spends half its time above 20 Hz, where the pulse turns rough.
The faintest: the 40 Hz gamma beat and the rising gamma session, mostly above 35 Hz, where listeners in one study noticed a beat half the time or less.

For what each band means in brain recordings, read the guides to delta, theta, alpha and gamma waves, and the stories behind the golden ratio and the Schumann resonance. Or press play at the top of the page: the 10 Hz alpha beat, 108 Hz in your left ear and 118 Hz in your right. Headphones on, volume low, and notice what you hear.

Sources and limits

Most binaural beat studies are small and short, and they differ in the beats they play, the tones they use and what they measure, so pooled numbers carry wide uncertainty. Brain recordings show that the brain processes the beat; they do not show that it moves the whole brain into a band. The hospital trials measured how people rated their own nerves. The edges in the diagram are approximate, as the reviews say, and no study has tested our recordings themselves.