The Best Music to Study To, Graded by the Evidence

Before the listHow the grades work

Each sound carries one of four badges. The badge describes the evidence, not how much we like the sound.

All 13 at a glance

The table is the article in one screen. The chart under it is the reason the article cannot end there.

Two bars pointing in opposite directions is about as honest a summary of this topic as you will get. There is no single best study sound, only a task and a person, and the answer moves when either one of them changes.

Does music help you study?

On average, no, and the average is misleading. Kämpfe, Sedlmeier and Renkewitz pooled 97 studies in 2011. Across 66 of them, covering 4,501 people, music against no music came out at r = .03. That is nothing. The authors say so themselves, and they explain why: the specific effects cancel each other out.

A 2022 systematic review by Cheah and colleagues went through 95 articles and 154 experiments and landed in the same place: a general detrimental effect of background music on memory and language tasks, with lyrics tending to be worse than instrumental. The harm showed up on difficult tasks rather than easy ones, which is the pattern you would expect if music is competing for something you need.

The explanation both sets of authors reach for is competition. Reading and verbal memory run on the same language machinery that processes the words in a song, so the song wins some of the time. Treat that as their interpretation rather than a settled mechanism. What is settled is the direction of the reading result.

Family 1The baseline: silence

Silence is the comparison condition in almost every study on this page, which makes it the scoreboard rather than a recommendation. Everything else gets measured against it, and on reading tasks very little comes out ahead.

The practical version costs about two pounds. Foam earplugs, or noise-cancelling headphones with nothing playing through them, or the floor of the library where talking is not allowed. If you have never tried reading a dense chapter in actual quiet, that is the experiment to run before you optimise a playlist.

Family 2Music

Six kinds. What separates the harmless ones from the costly ones comes down mostly to whether there are words in it, and how loud it is.

Souza and Barbosa put this to the test properly in 2023 in the Journal of Cognition, with University of Porto students. Their instrumental condition was a lo-fi hip hop playlist, the same genre as every study-with-me stream on YouTube. Against silence, across verbal recall, visual recall, reading and arithmetic, nothing they measured was credible in either direction.

So when people ask whether lo-fi is scientifically proven: it has been tested once, head to head with silence, and it neither helped nor hurt. The students in that study often felt it had helped them, which is the more interesting half of the result.

Vasilev, Hitching and Tyrrell ran four experiments at Bournemouth University in 2024, around 200 people each, measuring self-paced reading. Instrumental music against silence was essentially null there too, with slight speed-ups when the music was unfamiliar. Shih and colleagues found the same shape in 2012: their no-lyrics group moved from 109.3 to 106.7 on an attention test, which did not reach significance.

Instrumental music is safe for the same reason it is inert. With no words in it, there is nothing competing for the language processing you are reading with.

Kiss and Linnell gave 40 students a long, monotonous sustained-attention task and ran it twice: once in silence, once with music the students chose themselves. They asked people at intervals where their attention actually was.

The authors read this through the arousal curve: a boring task leaves you under-stimulated, and music you like tops you back up enough to stop the drift. That fits the data. It is also an explanation much easier to believe than to test.

In practice this is advice about admin. Formatting references, copying up notes, cutting flashcards, sorting a folder of past papers. Not reading. And not if your favourite songs have choruses you will end up mouthing along to.

Pietschnig and colleagues pooled around 40 studies in 2010. Mozart against no music came out at d = 0.37. Other music against no music came out at d = 0.38. Whatever is happening, Mozart is not the ingredient.

There is a detail here that people enjoy. Thompson and colleagues used the same Mozart sonata, K. 448, and found it damaged reading comprehension when it was played fast and loud. The piece that launched a thousand baby CDs is also the piece in the next section, on the wrong settings. So classical is not inherently safer than anything else. It tends to be played quietly, which is a different thing.

We searched for a peer-reviewed trial of game soundtracks or EDM mixes against silence on study-type tasks and could not find one. An undergraduate thesis turned up, which does not count. If someone tells you gamers’ music is engineered for focus so it must work, that is reasoning, not evidence.

You can still reason about it. Game music is written to loop without pulling your eye and usually carries no lyrics, which puts it in the same profile as lo-fi above. A lot of EDM is fast and loud, which puts it in the profile of the section below. Check the tempo, keep the volume down, and treat the result as a guess rather than a finding.

Thompson, Schellenberg and Letnic ran a 2×2 at the University of Toronto Mississauga: two tempos, two volumes, 25 undergraduates, Mozart’s K. 448 throughout. One cell fell apart.

The number you can act on is the volume. In this study the damage appeared at 150 bpm played at about 72 dB. Turn it down before you skip the track, because fast and soft was the best-performing cell of the four.

Small sample, one piece of music, a borrowed baseline. The claim this supports is a narrow one: fast and loud together hurt reading. Anything wider than that is not in the paper.

This is the finding that shows up everywhere you look. Souza and Barbosa measured verbal recall at d = −0.32 and visual recall at d = −0.33 against silence, both credible, and their participants correctly sensed that the lyrics were hurting them. Cheah and colleagues found the same tendency across 154 experiments. Vasilev and colleagues found lyrics slowed reading more than instrumental music did.

That last one comes with an honest caveat. The reading slowdown was tiny, d in the region of 0.04 to 0.07. Reading speed barely moves. What moves is how much of it you still have an hour later, which is the part you were revising for.

If you cannot give up vocals, there is one hedge worth trying: pick a language you do not speak. Sun and colleagues reported in 2024 that lyrics in the same language as the text did the most damage. We are quoting the direction and none of the numbers, because the cells in that study were around 15 people and the effect sizes are implausibly large. Treat it as may help, not as tested.

Family 3Noise and ambient sound

The results in this family come with a population attached, and the population usually matters more than the effect size does. Check who was in the study before acting on what it found.

Van Hedger and colleagues played nature sounds to 31 people and urban sounds to 32, then measured working memory with a backward digit span and a dual n-back. The nature group improved and the urban group did not.

The comparison was urban noise, not silence. That matters more than the effect size does. Nobody has shown that birdsong beats quiet, only that it beats traffic, so use nature sounds as break audio or as a mask for a noisy room rather than as an upgrade on a silent one.

The theory behind it is attention restoration: natural stimuli ask less of your directed attention, so it recovers. The study was testing that idea rather than confirming it.

This is Chart A at the top of the page. Nigg and colleagues at OHSU pooled 13 studies and 335 participants in 2024 and found g = 0.249, with a confidence interval running from 0.135 to 0.363 and a p-value below .0001. The interval is narrow and it does not cross zero, so the effect is small without being in much doubt.

It replicates further back. Söderlund, Sikström and Smart reported in 2007 that noise helped the ADHD group and worsened performance in controls. Helps and colleagues tested 90 children aged 8 to 10 in 2014 and found moderate white noise helped the sub-attentive children, left the average children unchanged, and hurt the most attentive ones. Three separate studies, and the same crossover shows up in all of them.

The proposed mechanism is stochastic resonance, packaged as the Moderate Brain Arousal model: a brain running low on arousal gets something useful out of background noise that a brain running at normal arousal does not. Proposed, not proven.

The comparison arm of the same meta-analysis: 11 studies, g = −0.212, confidence interval from −0.355 to −0.069, p = .0036. Helps and colleagues found the same thing in the children who were already good at paying attention.

This is the most counterintuitive result on the page, and it runs directly against the advice of roughly every productivity video on the internet. If you do not struggle with attention, the pooled evidence says white or pink noise ↓ makes it slightly worse.

One exception, offered as reasoning rather than as a tested claim: if your room is loud and unpredictable, steady noise may still beat intermittent chatter. At that point you are not choosing between noise and quiet, only between one kind of noise and another.

Mehta, Zhu and Cheema ran five experiments in 2012 using everyday multi-talker background noise. At 70 dB people did better on creative tasks than at 50 dB. At 85 dB they did worse. The authors read it as moderate noise making processing slightly harder, which nudges people toward more abstract thinking, while loud noise simply swamps them.

This is one paper with five experiments inside it. We searched for replications and turned up nothing in either direction, which is not the same as saying nobody has tried.

So the café is for thinking up essay angles, not for learning the content of them. The finding does not travel between the two.

In plain EnglishWhat is white or pink noise?

Sound is made of high notes and low notes. White noise is all of them at once, at the same loudness, all the time. It comes out as a flat hiss. An untuned radio, a fan, a shower running in the next room.

Pink noise is the same thing with the high notes turned down. Softer, deeper, less hissy. Rain on a roof rather than static.

Neither one is music, and that is the entire point of them. There is no tune to follow, no beat to tap and no words to catch, so there is nothing in the sound for your attention to snag on. It is closer to a blanket thrown over the room than to something you listen to.

You will also see brown noise, which goes another step in the same direction: deeper again, more like a distant waterfall. It was not part of the meta-analysis above, so this page does not grade it.

Family 4Brain hacks

Garcia-Argibay, Santed and Reales pooled 22 studies and 35 effect sizes in 2019 and reported g = 0.45. That is the number every binaural beats page quotes, and it is real. What gets left out is that the pool mixes memory, attention, anxiety and pain, and that the authors themselves say the direction and magnitude depend on the frequency used, how long you listen and when you listen. Exposure before a task beat exposure during it.

Then there is the mechanism. Ingendoh, Posny and Heine reviewed 14 studies in 2023 to see whether binaural beats actually entrain brainwaves. Five supported it, eight contradicted it, one was mixed. The explanation everyone repeats is the part that has not held up.

Melnichuk and colleagues went finer in 2025, testing 16 parameter combinations on 64 analysed participants. Exactly one improved general attention: gamma beats over a low carrier tone with white noise underneath. None of them stopped attention sliding over time. Your app’s focus preset is unlikely to be that combination, and it will not tell you what it is.

We have not drawn a chart here on purpose. Three papers, three incompatible outcome measures, and putting them on one axis would imply a comparison that does not exist.

Rauscher, Shaw and Ky published a one-page paper in Nature in 1993. College students did better on a spatial task after hearing Mozart’s K. 448. Rauscher and Hinton later wrote that the effect lasted up to about 15 minutes and that the claim was never about general intelligence in the first place.

Chabris put a critique in Nature in 1999: any enhancement is small and does not reflect any change in IQ or reasoning ability in general. Then Pietschnig and colleagues pooled nearly 40 studies and more than 3,000 participants in 2010 and found little evidence left for a specific Mozart effect.

Worth saying the fair version too. The meta-analysis did not find nothing. It found a small effect that any music produces, on spatial tasks, for about a quarter of an hour. That is a long way from the claim that got printed on CD sleeves.

Five mistakes worth stopping

• Reading with lyrics on. It is the best-replicated harm on this page, and the cost lands on recall rather than on how fast the page goes by.
• Turning it up to drown out the room. Fast and loud was the one setting that sank reading scores. Earplugs first, then quiet sound if you still want it.
• Choosing music for the transitions between tracks. That advice traces back to a 2007 Stanford scan of people lying still and listening to symphonies. Nobody in that study was studying anything, and a moment where your brain perks up at a change in the music is a moment it is not on your notes.
• White noise because a productivity video said so. Unless you struggle with attention, the pooled evidence points the other way.
• Playing Mozart to get smarter. Up to about 15 minutes, spatial tasks only, and other music matched it.

A sixth, from experience rather than from a paper: the free YouTube compilations that run an ad every twenty minutes are worse than silence for reasons no study needs to establish.

What to play, by task

Everything above, rearranged around what you are actually sitting down to do.

Then run your own test, because none of these effect sizes are big enough to be obvious from the inside. Pick one task, alternate silence and your playlist across a week, and count something real: pages read, cards learned, questions finished. Not how focused you felt.

That last distinction is the whole article. In the Porto study the students felt the lo-fi was helping. It was not. The same gap between feeling productive and being productive runs through our ranking of study methods by effect size, where the two techniques students rate most highly are the two the evidence says to drop.

Questions people actually ask

Sources

Every link goes to the paper, its publisher record, or a copy hosted by one of its authors. None of them goes to a blog summarising it.

1. 01Kämpfe, J., Sedlmeier, P., & Renkewitz, F. (2011). The impact of background music on adult listeners: a meta-analysis. Psychology of Music, 39(4), 424–448. doi
2. 02Cheah, Y., Wong, H. K., Spitzer, M., & Coutinho, E. (2022). Background music and cognitive task performance: a systematic review of task, music and population impact. Music & Science, 5. doi
3. 03Souza, A. S., & Barbosa, L. C. L. (2023). Should we turn off the music? Music with lyrics interferes with cognitive tasks. Journal of Cognition, 6(1), 24. doi
4. 04Vasilev, M. R., Hitching, L., & Tyrrell, S. (2024). How does background music during reading influence text processing? Journal of Cognitive Psychology, 36(1). doi
5. 05Shih, Y.-N., Huang, R.-H., & Chiang, H.-Y. (2012). Background music: effects on attention performance. Work, 42, 573–578. doi
6. 06Thompson, W. F., Schellenberg, E. G., & Letnic, A. K. (2012). Fast and loud background music disrupts reading comprehension. Psychology of Music, 40(6), 700–708. doi
7. 07Kiss, L., & Linnell, K. J. (2021). The effect of preferred background music on task-focus in sustained attention. Psychological Research, 85(6), 2313–2325. doi
8. 08Perham, N., & Currie, H. (2014). Does listening to preferred music improve reading comprehension performance? Applied Cognitive Psychology, 28(2), 279–284. doi — cited for direction only; we could reach the publisher record but not the abstract, so no figures from it are quoted here.
9. 09Sun, Y., et al. (2024). The effect of background music with lyrics on reading comprehension. Frontiers in Psychology, 15, 1363562. doi — direction only; the cells are small and the reported effect sizes are implausibly large.
10. 10Nigg, J. T., et al. (2024). Evidence for white noise as a possible therapeutic modality for ADHD: a systematic review and meta-analysis. Journal of the American Academy of Child & Adolescent Psychiatry, 63(8), 778–788. doi
11. 11Söderlund, G., Sikström, S., & Smart, A. (2007). Listen to the noise: noise is beneficial for cognitive performance in ADHD. Journal of Child Psychology and Psychiatry, 48(8), 840–847. doi
12. 12Helps, S. K., Bamford, S., Sonuga-Barke, E. J. S., & Söderlund, G. B. W. (2014). Different effects of adding white noise on cognitive performance of sub-, normal and super-attentive school children. PLOS ONE, 9(11), e112768. doi
13. 13Van Hedger, S. C., Nusbaum, H. C., Clohisy, L., Jaeggi, S. M., Buschkuehl, M., & Berman, M. G. (2019). Of cricket chirps and car horns: the effect of nature sounds on cognitive performance. Psychonomic Bulletin & Review, 26(2), 522–530. doi
14. 14Mehta, R., Zhu, R., & Cheema, A. (2012). Is noise always bad? Exploring the effects of ambient noise on creative cognition. Journal of Consumer Research, 39(4), 784–799. doi
15. 15Garcia-Argibay, M., Santed, M. A., & Reales, J. M. (2019). Efficacy of binaural auditory beats in cognition, anxiety and pain perception: a meta-analysis. Psychological Research, 83, 357–372. doi
16. 16Ingendoh, R. M., Posny, E. S., & Heine, A. (2023). Binaural beats to entrain the brain? A systematic review. PLOS ONE, 18(5), e0286023. doi
17. 17Melnichuk, A., Cooper, S., & Hawk, L. (2025). Effects of binaural beat parameters on sustained attention. Scientific Reports, 15, 4308. doi
18. 18Rauscher, F. H., Shaw, G. L., & Ky, K. N. (1993). Music and spatial task performance. Nature, 365, 611. doi
19. 19Rauscher, F. H., & Hinton, S. C. (2006). The Mozart effect: music listening is not music instruction. Educational Psychologist, 41(4), 233–238. doi
20. 20Chabris, C. F. (1999). Prelude or requiem for the Mozart effect? Nature, 400, 826–827. doi
21. 21Pietschnig, J., Voracek, M., & Formann, A. K. (2010). Mozart effect–Shmozart effect: a meta-analysis. Intelligence, 38(3), 314–323. doi
22. 22Sridharan, D., Levitin, D. J., Chafe, C. H., Berger, J., & Menon, V. (2007). Neural dynamics of event segmentation in music. Neuron, 55(3), 521–532. doi — the Stanford scanning study that gets quoted as proof music aids concentration. It measured neither.

Reviewed every twelve months, and sooner if a new meta-analysis lands on background music, lo-fi or binaural beats. Last checked 17 September 2026.