Binaural Beats for Sleep: How They Work and How to Use Them
Written by the Binaura Team · Published 2026-08-12 · Last reviewed 2026-08-12
What happens when each ear hears a slightly different tone, and a three-step evening routine you can run tonight. What you will not find here is a number for how much faster you will fall asleep, because there is no honest one to give.
What happens when each ear hears a slightly different tone
Play a steady 200 Hz tone into your left ear and a steady 204 Hz tone into your right, and you do not hear two tones. You hear one, and it pulses — four times a second, because the beat frequency is the difference between the two tones (Oster, Scientific American, 1973).
The pulse is the interesting part, because it is not in the air. Neither earpiece is producing a 4 Hz sound, and 4 Hz is in any case far below the 20 Hz bottom of human hearing (Neuroanatomy, Auditory Pathway, StatPearls); nothing in the room is wobbling four times a second. The two tones never meet acoustically at all. They meet in your hearing, where the brainstem compares the signal arriving at one ear with the signal arriving at the other — the same comparison it uses to tell you which direction a sound came from. The beat is what that comparison sounds like when the two inputs refuse to stay in step.
That is why the effect has limits you can hear for yourself. The biophysicist Gerald Oster, whose 1973 Scientific American article is the paper most of this field still cites, reported that binaural beats are clearest with a carrier tone of about 440 Hz, become less distinct above that, and vanish above roughly 1,000 Hz — no listener he tested heard beats from tones above 900 Hz, though he noted other investigators detecting them up to almost 1,500 Hz, and that below about 90 Hz the beat becomes elusive again (Oster, Auditory beats in the brain, Scientific American 229(4), 1973; the carrier limit is restated in a 2024 review of binaural beats and the brain). A binaural track is therefore built from low, quiet tones by necessity, not by taste.
It also means the delivery is not optional. Two tones cannot arrive separately at two ears through one speaker, and a mono mix has already summed them before they leave the file. Do you need headphones for binaural beats? works through what breaks it — speakers, one earbud, a call profile — and which headphone features are marketing.
Where the mechanism stops and the claim starts
Almost every sleep track you will find is labelled with a brainwave band. Delta is the one attached to sleep, and clinical EEG puts it at 0.5–4 Hz (Normal EEG Waveforms, StatPearls, NCBI Bookshelf, which lists δ 0.5–4 Hz, θ 4–7 Hz, α 8–12 Hz and β 13–30 Hz).
Here is the honest reading of that label: a “delta track” is named after the band. It is a track whose two tones differ by a number that falls inside the delta range. That is a fact about the audio file. Whether listening to it makes your own EEG do anything delta-shaped is a separate question, and it is the question the research has struggled with for fifty years. Research is mixed, and most studies are small. Our evidence page sets out that literature study by study — sample sizes, what each one measured, and the null results alongside the positive ones — including a double-blind sleep study in which both the binaural group and the sham group improved.
So the useful frame is not “this will put my brain into delta.” It is “this is a steady, low, undemanding sound, and I am choosing it deliberately at the end of the day.” That is a smaller claim. It is also one that survives contact with the evidence.
What a night actually looks like, and why it changes the routine
Adult sleep is not one flat state. Deep sleep — stage N3, the slow-wave stage delta is the signature of — makes up roughly 10–20% of total sleep time and falls mostly in the first third of the night; N1 accounts for about 2–5%, N2 for 45–55%, and REM for 20–25% (Normal Sleep, Sleep Physiology, and Sleep Deprivation, Medscape). The cycles themselves run 70–100 minutes early in the night and lengthen to 90–120 minutes later.
Two practical consequences follow, and neither requires believing anything about entrainment. The first is that the part of the night audio can plausibly touch is the beginning — the stretch before you are asleep and the first cycle after. The second is that a track running all night is not doing eight hours of work; it is doing twenty minutes of work and then playing to a sleeping person, at whatever volume you left it at.
A three-step evening routine
1. Decide when it ends before it starts
Pick the length first, and pick it short. A single track that stops on its own is better than a playlist that does not, because a playlist makes the decision to stop your problem, and you will be asleep or nearly asleep when it arrives. Given that deep sleep concentrates in the first third of the night (Medscape, above), audio that covers the transition and then gets out of the way is matched to what the night is doing.
If you want a number to start from, use the one you would use for a focus session and shorten it. How to listen to binaural beats covers session length as a practice rather than a prescription.
2. Set the volume while you are still awake
This is the step people skip, and it is the only step on this page with a real safety argument behind it. A steady low tone is undramatic, which makes it easy to creep upward on, and sleep audio runs for a long time. The WHO safe-listening guidance treats 80 dB as safe for up to 40 hours a week, with the safe duration falling to about four hours a week at 90 dB — figures about audio exposure in general, not about this genre. Our page on whether binaural beats are safe states that guidance in full, along with the epilepsy caution.
Set it quiet enough that you could hear someone say your name over it. Then leave it.
3. Let the track finish without you
The last step is to stop managing it. No checking whether it is “working,” no switching tracks because this one feels wrong, no reaching for the phone to change the frequency. Every one of those is a reason to be awake, and being awake is the thing you are trying to stop doing.
If a track is uncomfortable, that is different — stop it and try something else another night. There is no dose to reach here and nothing to push through toward.
What this is not
Binaura offers soundscapes and breathing exercises. It is not a treatment for insomnia, anxiety or any other condition, it does not diagnose anything, and it is not a substitute for care from a clinician or for anything a clinician has asked you to do. No study on our evidence page evaluated Binaura as a product.
If sleep or anxiety problems keep happening or interfere with daily life, talk with a licensed clinician.
Sources
- Oster, G. (1973). Auditory beats in the brain. Scientific American 229(4), 94–102. Carrier-frequency limits for the perception of binaural beats. Limitation: an individual investigator's observations from 1973, reported as ranges, with the author noting that intensity and earphone type change the result.
- A review of binaural beats and the brain (2024), PMC. Restates the sub-1,000 Hz carrier limit and the brainstem account. Limitation: a narrative review, not a quantitative synthesis.
- Normal EEG Waveforms, StatPearls, NCBI Bookshelf. Clinical EEG frequency bands. Limitation: band edges differ between references, as the next source shows.
- Normal Sleep, Sleep Physiology, and Sleep Deprivation, Medscape. Sleep-stage proportions and cycle lengths in adults. Limitation: population averages that vary with age and between individuals; they describe sleep, not the effect of any audio on it.
More on the setup and the practice: whether headphones are required, how to listen, and what the frequency numbers mean. The band itself is unpacked in delta waves explained, and the research in what the evidence says. Short answers are in the FAQ.
How this post was written: it is built from published research rather than from house opinion, and every study behind a number on this page is linked in place, next to the figure it supports, with the limitation the authors themselves stated. If a claim here has no link, it is not a claim we are making. Corrections go to Binaura support.