Sounds for Deep Work: Noise, Beats or Music?

When you sit down to do something that needs your full attention, there are three kinds of sound on offer: steady noise, music, or a slow beat made for the purpose. They are very different sounds, and the research on each is thinner than the buzz around them suggests. Here is what has been studied, what hasn’t, and a simple way to find what suits you.

A note on labels. We rate the evidence for what people link with a sound, not the sound itself, using three labels: Research Supported, Emerging / Mixed and Traditional / Unproven. When we’re unsure, we use the lower one.

Steady noise: the level and the task matter

Steady noise is a wash of sound with no words or melody in it. The most useful finding is that there is no single rule.

One set of five experiments (Mehta, Zhu and Cheema, 2012) compared three room noise levels: 50 dB, 70 dB and 85 dB. People did better on creative tasks at the moderate level, 70 dB, than at the low one, and the loud level, 85 dB, reversed that. The authors suggest moderate noise makes thinking slightly harder, which nudges people toward more abstract ideas. Two limits are worth keeping in mind. These were laboratory rooms and creative tasks, which may not match your task. And the levels are not a target for headphones. Keep your own volume low and comfortable.

A much larger review (Szalma and Hancock, 2011) pooled 242 studies and 797 results. It found that three things shape how noise affects performance: whether the noise is continuous or intermittent, what kind of noise it is, and what task someone is doing. The two common explanations, that noise adds alertness or that it covers up distractions, each got mixed support.

There is also a small result for plain white noise. In a laboratory study of healthy adults (Rausch, Bauch and Bunzeck, 2014), white noise played while people looked at pictures slightly raised their later recognition memory, while a plain tone did not. The authors called the effect slight, and it covers one kind of memory task, not a working day.

Evidence label: Emerging / Mixed. There are real studies, but they are lab-based, they use different tasks and levels, and they point different ways.

Brown noise: a favorite with little direct research

Brown noise is a steady sound with most of its energy in the low pitches. Its power falls by 6 dB with each octave you go up, so it comes across as a soft low roar, a bit like a waterfall, heavy rain or distant thunder. The name comes from Robert Brown and Brownian motion, not from the color.

It is often suggested online as a sound for studying. We could not find a study of brown noise and concentration in general adult listeners that we were able to read and verify. So we can describe the sound, and we can say it is often suggested, but we cannot point to research that shows it does anything in particular.

Evidence label: Traditional / Unproven. That is not a verdict against it. It only means the idea rests on personal reports.

Music: the words are the main question

Music is the best-studied of the three, mostly through reading. A review of 65 studies on auditory distraction during reading (Vasilev, Kirkby and Angele, 2018) compared reading in silence with reading alongside noise, speech or music. The authors conclude that all three made comprehension slightly worse, by small amounts.

The clearer pattern was about words. Music with lyrics was more distracting than music without, and speech you can understand was about as distracting as lyrics. Instrumental music was close to no effect and not clearly different from plain noise. The authors note that some of these comparisons rest on few studies, and that they compare results across studies rather than testing everyone in the same way.

The practical reading: when your work involves reading or writing, sound with words in it competes for the same attention. For other kinds of work, we did not find equally strong evidence, so your own experience is a fair guide.

Evidence label: Research Supported, for the narrow idea that words in the background make reading a little harder. It does not say music is good or bad for work in general.

Beats: a slow pulse in the sound

A binaural beat is made from two steady tones, one in each ear, that differ slightly. A 200 Hz tone in your left ear and a 210 Hz tone in your right ear are heard as a 10 Hz beat. Because each ear needs its own tone, binaural beats need stereo headphones. On a phone speaker the same idea is played as a pulsing tone instead, where the volume of one tone rises and falls.

Two meta-analyses have looked at this. The more recent one (Basu and Banerjee, 2023) pooled 15 studies on memory and attention tasks and found a medium overall effect. Looking at the individual studies, though, the results conflicted, especially for beats in the theta and beta ranges. The authors describe the support as modest and ask for more careful studies. An earlier meta-analysis (Garcia-Argibay and colleagues, 2019) of 22 studies across several outcomes also found a medium overall effect, and noted that it depended on the beat frequency, how long people listened, and whether the beat started before the task.

Pooled results from small studies with different methods are a starting point, not a settled answer.

Evidence label: Emerging / Mixed.

Try it yourself

You don’t need a study to learn what works for you. A simple test:

Pick one sound and a task you do often.
Keep the volume low, quiet enough that you could still hear someone speak softly.
Work with it for a few sessions, then work in silence for a few more.
Notice which felt easier to settle into.

One or two sessions tell you very little, so give each sound several tries before deciding. If you can’t tell a difference, silence is a perfectly good choice. OscillaMind, which is coming to Android first, is being built so you can try brown noise, rain and beats side by side at a low volume.

OscillaMind is for listening and routines, not medical care. If you have a health concern, talk to a doctor.

FAQ

Is brown noise better than white noise for work? We could not find a verified study that compares them for general adult listeners. They differ in shape: white noise is bright and hissy, brown noise is deep and soft. Most people choose by which one they find easier to ignore.

Should I listen to music with lyrics while I work? For reading, studies found lyrics are more distracting than music without words. If your task involves words, an instrumental track or steady noise is the safer choice. For other tasks, try both and see.

Do binaural beats work for concentration? The evidence is mixed. Meta-analyses report a medium average effect, but the individual results conflict, and the studies are small and varied. If you try them, use stereo headphones at a low volume.

How loud should work sound be? Low and comfortable. The loudness figures in studies come from laboratory rooms, not from headphones. A good rule is that you could still hear a quiet voice over it.

Sources

Mehta, Zhu & Cheema (2012), Is Noise Always Bad? Exploring the Effects of Ambient Noise on Creative Cognition, Journal of Consumer Research. https://academic.oup.com/jcr/article/39/4/784/1795620
Szalma & Hancock (2011), Noise Effects on Human Performance: A Meta-Analytic Synthesis, Psychological Bulletin. https://peterhancock.ucf.edu/noise-and-human-performance-a-meta-analytic-synthesis/
Rausch, Bauch & Bunzeck (2014), a laboratory study of white noise and recognition memory, Journal of Cognitive Neuroscience. https://pubmed.ncbi.nlm.nih.gov/24345178/
Vasilev, Kirkby & Angele (2018), Auditory Distraction During Reading: A Bayesian Meta-Analysis of a Continuing Controversy, Perspectives on Psychological Science. https://pmc.ncbi.nlm.nih.gov/articles/PMC6139986/
Basu & Banerjee (2023), a meta-analysis and systematic review of binaural beats, memory and attention, Psychological Research. https://pubmed.ncbi.nlm.nih.gov/35842538/
Garcia-Argibay, Santed & Reales (2019), a meta-analysis of binaural auditory beats across several outcomes, Psychological Research. https://pubmed.ncbi.nlm.nih.gov/30073406/
Brownian noise (reference entry). https://en.wikipedia.org/wiki/Brownian_noise