Does music help with sleep? What the trials show

What did the trials of music and sleep find?

The fullest answer comes from a Cochrane review by Jespersen and colleagues, updated in 2022. It gathered 13 randomised trials with 1,007 adults who had trouble sleeping. In every trial people listened to recorded music once a day, for 25 to 60 minutes and 36 on average, over anything from three days to three months, and most were asked to listen at bedtime. Their nights were compared with those of people who had no music, or only the usual care, such as lessons in good sleep habits.

The main measure was a sleep questionnaire scored from 0 to 21, where a higher score means poorer sleep. Pooled across 10 trials and 708 people, the music groups ended 2.79 points lower than the groups without music, with a 95 per cent confidence interval of 1.72 to 3.86 points. The reviewers rate that evidence as moderate certainty, and put the size of the change at about one standard deviation: a large shift in how well people said they slept.

Points lower on the 0 to 21 scale with nightly music

Group	Average	95% range
All trials: 10 trials, 708 people	2.79	1.72 to 3.86
8 to 21 days of music: 5 trials, 343 people	2.24	1.58 to 2.90
22 days or more: 5 trials, 365 people	3.36	1.10 to 5.63
Researchers chose the music: 5 trials, 370 people	3.31	1.29 to 5.32
Listeners chose from a set: 5 trials, 338 people	2.33	1.29 to 3.37
Music on its own: 8 trials, 583 people	2.85	1.51 to 4.18
Music with relaxation instructions: 2 trials, 125 people	2.64	1.54 to 3.74

Download the trial results (CSV)

The subgroups agree with each other. Music for 8 to 21 days and music for three weeks or more, music the researchers picked and music the listeners chose from a set, music alone and music with relaxation instructions: the review found no clear difference within any of those pairs. The trials did differ a good deal in how large a change each one found, which the review measures as heterogeneity of 81 per cent, so the 2.79 is an average over quite different results, not a figure to expect for any one person. None of the 13 trials reported any harm from the music.

What does a 2.8 point change on a sleep questionnaire mean?

The questionnaire is the Pittsburgh Sleep Quality Index, published by Buysse and colleagues in 1989. It asks 19 questions about the past month. The answers feed seven parts, each scored from 0 to 3, which add up to one score from 0 to 21, as Trahan and colleagues set out. In Buysse's first study a score above 5 sorted poor sleepers from good ones well, picking out 89.6 per cent of the poor sleepers and 86.5 per cent of the good sleepers correctly.

The seven parts of the score

Part	What it asks about
Subjective sleep quality	How well you would say you slept
Sleep latency	How long it took to fall asleep
Sleep duration	How many hours of actual sleep
Habitual sleep efficiency	Hours asleep against hours in bed
Sleep disturbances	Waking in the night, being too hot or cold, and other interruptions
Use of sleeping medication	How often anything was taken to get to sleep
Daytime dysfunction	Trouble staying awake or keeping up enthusiasm in the day

So the trials' average, 2.79 points, is about 13 per cent of the whole scale. Every point of it comes from the sleeper's own answers. That is the questionnaire's strength, since part of a good night is how it feels, and also its limit: a self-rated score cannot separate a change in sleep from a change in how people judge their sleep. The Cochrane reviewers judged all 13 trials at high risk of performance bias, because nobody can be blind to whether they are listening to music.

Do sleep sensors show the same change?

Not so far. Three of the trials, with 136 people, also measured sleep with instruments, and on those measures the review found that music may not change how long people took to fall asleep, how long they slept, how much of their time in bed they spent asleep, or how often they woke. The surest of those instruments is polysomnography, the overnight recording of brain activity, breathing and movement, which the review calls the gold standard of sleep assessment.

The clearest single example is a trial by Jespersen and colleagues of 57 adults with long-standing sleep problems. They were split three ways, into music at bedtime, an audiobook at bedtime, or a waiting list, and their sleep was recorded with polysomnography and with actigraphy, a movement sensor worn on the wrist. The authors report "a significant effect of the music intervention on perceived sleep improvement and quality of life, but no changes in the objective measures of sleep."

An audiobook is a useful comparison, because it is also something pleasant to listen to in bed. In a trial of 94 students who slept poorly, Harmat and colleagues gave one group 45 minutes of relaxing classical music at bedtime for three weeks, another an audiobook, and a third nothing. Only the music group's questionnaire scores moved: in the authors' words, sleep quality "did not improve statistically significantly in the audiobook and the control group". One small trial cannot settle why, but it suggests that something about music, not only lying still and listening, made the difference to how those students rated their nights.

What kind of music did the trials use?

Mostly slow, steady and calm. Four trials in the review described the music they played, and it shared a low tempo, from 52 to 85 beats a minute, a stable dynamic structure and no strong rhythmic accents. The genres ran wide: Western and Chinese classical music, Buddhist songs, new age, lullabies, Persian traditional music, Chinese Five Elements tone music, ambient, popular oldies and jazz.

Who chose it did not seem to matter much. In some trials the researchers chose the music; in others, such as Lai and Good's trial of 60 people aged 60 to 83, listeners chose among six 45-minute tapes of calm music, five Western and one Chinese. Both kinds of trial show the change, and the review found no clear difference between them. None of the music the review describes was a single tone at a named frequency.

Outside the trials, people's own choices look much the same. Trahan and colleagues asked 651 people about music and sleep: 62 per cent said they used music for sleep, and they named 545 artists across 14 genres, with Johann Sebastian Bach mentioned most, by 15 people, ahead of Ed Sheeran and Mozart. Scarratt and colleagues measured 985 sleep playlists on Spotify, 225,626 tracks, against more than 3.7 million tracks of everyday listening. Sleep music was softer and slower, more often instrumental and more often played on acoustic instruments. Its average loudness on Spotify's scale was -19.78, against -9.6 for everyday music, and its average tempo 104.93 beats a minute, against 120.07. Yet three of the six clusters they found included popular tracks that were faster, louder and more energetic than the average sleep track.

Why might music make a difference at bedtime?

Researchers have offered six reasons, and they are not equally well supported. Dickson and Schubert sorted them in a review of the literature:

• Masking, music covering unwanted background noise, had support.
• Relaxation, distraction from stressful thoughts, and enjoyment of music you like had mixed support.
• Expectation, what a listener and their culture believe about music, had possible support.
• Entrainment, body rhythms falling into step with the beat of the music, had mixed possible support.

Listeners give much the same reasons. The people in Trahan's survey named four: something in the music itself, the habit of a bedtime routine, a state of body or mind that suits sleep, and blocking out sounds and thoughts that would otherwise keep them awake. And under Harmonance's most-watched video, a listener wrote:

I was able to meditate peacefully without outside noise and distraction. Thank you for sharing

That is masking, put plainly. Another listener, under the same video, wrote of falling asleep to it and waking with the television still on, so a long, steady sound has to suit the whole night, not only the first minutes.

Can music keep you awake?

It can, mostly through earworms, the tune that keeps replaying in your head. Scullin and colleagues surveyed 199 adults and found that people who listen to music often reported persistent earworms at night, which went with worse sleep quality. They then brought 50 young adults into a sleep laboratory and played them popular songs before bed, either with their lyrics or in instrumental versions. Against what many people would guess, the instrumental versions brought more earworms at night and worse sleep quality as polysomnography measured it. Catchy, familiar music can follow you into bed.

Interruptions matter as well. Comments about adverts breaking into the sound come up again and again under Harmonance's videos, and one puts it in a line:

You shouldn't put ads between meditation music!

That is why the full versions play free and without adverts on harmonance.com. Not everyone takes to a steady tone, either. One listener stopped after nine minutes and asked whether it was normal to feel unwell. A tone that does not suit you is a reason to switch it off, and no tone is worth sitting through.

Volume is the last thing to watch over a whole night. The US National Institute on Deafness and Other Communication Disorders says sounds at or below 70 dBA, "even after long exposure, are unlikely to cause hearing loss", and that long or repeated exposure at 85 dBA or more can cause it. Music at bedtime needs far less than either: just enough to hear.

Is there a best frequency for sleep?

None of the research here points to one. The music trials played ordinary recorded music, and none of them tested one tone against another. Three lines of research come closest, and none of them names a best frequency.

Binaural beats. A binaural beat plays a slightly different tone in each ear, and the brain hears the difference as a slow pulse. In a trial by Bang and colleagues, 43 adults with mild sleep trouble finished two weeks of 30 minutes of music before sleep, with or without a theta binaural beat mixed in. Both groups reported fewer sleep problems, with no clear difference between them, and the authors conclude that the beat with music "does not induce evident improvement of sleep disturbances more than does pure music". On whether the brain follows a beat at all, a systematic review by Ingendoh and colleagues found 14 studies: five reported results in line with the idea, eight did not, and one was mixed. The guides to whether binaural beats work and to theta waves go further.

Sound timed to deep sleep. In a sleep laboratory, Ngo and colleagues played short bursts of sound, timed by a computer watching each sleeper's brain recording, to land on the peaks of the slow waves of deep sleep. Sound in step with those waves strengthened them; the same sound out of step "remained ineffective". The timing came from a live recording of each person's own brain, which a recording played from a speaker cannot do. The guide to delta waves explains the slow waves themselves.

Noise. Steady noise, such as white noise, is widely used at night. A systematic review by Riedy and colleagues of 38 studies rated the evidence for continuous noise as a sleep aid as very low quality, "which contradicts its widespread use", and warned that it "may also negatively affect sleep and hearing".

The US National Center for Complementary and Integrative Health reads the evidence the same way: moderate-certainty evidence that listening to music has a beneficial effect on how people rate their sleep, and mixed results in a review of 16 studies of music for sleep in older adults. That evidence is about music in general. It says nothing about any one pitch, the Moon tone included.

What does the Can't sleep page play?

The Can't sleep page leads with the Moon tone, 210.42 Hz, followed by Neptune at 211.44 Hz and 963 Hz. The Moon tone comes from Hans Cousto's cosmic octave. Cousto's table takes the synodic month, new moon to new moon, as 29.530588 days, and NASA's Moon fact sheet gives the same cycle as 29.53 days. Read as one very slow vibration, a month of 2,551,442.8 seconds is 0.000000392 cycles a second. Doubled 29 times, which multiplies it by 536,870,912, it reaches 210.42 Hz, close to a G sharp. NASA's rounded figure gives 210.423 Hz, 0.004 Hz higher. Had Cousto used the Moon's orbit against the stars, 27.3217 days, the same 29 doublings would give 227.43 Hz instead. The guide to planetary frequencies sets out the whole method, and the Moon sits with the other planets in the cosmic octave collection.

The trials point to slow, steady, quiet sound, so Harmonance measured how steady the page's own sound is. Every Moon player plays a 100-second recording of the tone on a loop, and the Strings mode plays a 180-second arrangement of slow strings. Measured second by second with the broadcast loudness meter, EBU R 128, the plain tone holds at -22.9 LUFS and never moves more than 0.7 LU, or more than 0.2 LU from one second to the next. Its strongest overtone, at 420.84 Hz, sits 23.1 dB below it, so almost all of its energy is in the single pitch.

The page's two sounds, measured

Measure	Moon tone	Strings mode
Loudness overall	-22.9 LUFS	-22.9 LUFS
Loudness range	0.7 LU	3.5 LU
Quietest and loudest second	-23.3 and -22.6 LUFS	-29.9 and -20.7 LUFS
Largest change in one second	0.2 LU	4.0 LU
Strongest pitch	210.42 Hz	210.81 Hz

Download the loudness readings (CSV)

The strings move more, as music does. At the page's own level they average the same -22.9 LUFS as the tone, and their strongest pitch, 210.81 Hz, sits within a few cents of it. But their loudness range is 3.5 LU, and five times in each three minutes the sound thins and dips below its average, four times by 5 to 7 LU and once by about 3 LU, before it swells back. Neither is a sound any trial tested. Both sit close to what the trials describe: no beat to follow, no words to catch, and a level that changes slowly or not at all.

If you would rather hear a beat that slows down through the night, the Falling Delta Session is a nine-hour binaural beat for headphones that falls from 3.5 to 1.5 beats a second, once, then stops, and the binaural beats collection holds the other sessions. The Moon tone's own page has the tone on its own, with the story of where it comes from.

How do you try music at bedtime?

• Give it a few weeks. The trials ran from three days to three months, and the change showed both in trials of 8 to 21 days and in longer ones.
• Keep it short and quiet. Trial sessions lasted 25 to 60 minutes, 36 on average. Keep the level just loud enough to hear, and let a timer end the sound: the guide to Harmonance Studio shows how to set one.
• Choose slow and steady. The trials' music was slow, with no strong beat and few changes in loudness. If a tune keeps replaying in your head, try something with less of a tune to it, such as a single tone.
• Notice for yourself. Try a week with music and a week without, and note each morning how the night felt. The trials' questionnaire asked much the same: how well you would say you slept.

The Can't sleep page plays the Moon tone, Neptune and 963 Hz in one place, Harmonance Studio adds timers and ambient layers for a session of your own, and the guide to what sound healing research shows looks at the wider evidence.

Sources and limits

The figures for music and sleep come from one Cochrane review and the trials it pooled, which used ordinary recorded music and mostly measured how people rated their own sleep. The few that measured sleep with instruments found no clear change. No study has tested the Moon tone, a Harmonance tone or any single pitch for sleep. The loudness readings come from the site's own audio files, measured by Harmonance: the strings at the page's default level, without the quiet room sound the page adds beneath them. How loud either is in your room depends on your device and its volume. The listener comments are a few people's words, quoted whole and without names, not findings.

1. Jespersen and colleagues, music for adults with poor sleep, the 2022 Cochrane review: 13 trials, 1,007 adults.
2. Buysse and colleagues, the Pittsburgh Sleep Quality Index: the questionnaire and its cut-off.
3. Trahan and colleagues, music and sleep in an online survey of 651 people.
4. Scarratt and colleagues, the audio features of 985 sleep playlists.
5. Jespersen and colleagues, a trial of bedtime music against an audiobook: questionnaires and sensors.
6. Harmat and colleagues, music, an audiobook or nothing for 94 students.
7. Lai and Good, calm music at bedtime for 60 older adults.
8. Dickson and Schubert, how does music aid sleep?: six proposed reasons, weighed.
9. Scullin and colleagues, bedtime music, earworms and sleep.
10. Bang and colleagues, music with and without a theta binaural beat.
11. Ingendoh and colleagues, binaural beats to entrain the brain?: a systematic review.
12. Ngo and colleagues, sound timed to the slow waves of deep sleep.
13. Riedy and colleagues, noise as a sleep aid: a systematic review.
14. NCCIH, music and health: what you need to know.
15. NIDCD, noise-induced hearing loss: safe listening levels.
16. Cousto, tuning data for the Earth, Moon, planets and Sun: the Moon tone.
17. NASA, Moon fact sheet: the synodic and sidereal months.