The science, with the weak parts left in

Nothing in Hum rates strong. The best evidence is moderate, and it’s about music, noise or breathing in general, not Hum’s own sound. Hum isn’t a treatment. Here are the numbers, so you can judge for yourself.

Researched 4 November 2026. Not medical advice.

The numbers

Each line shows a pooled result and its 95% confidence interval. We’ve flipped the results where a lower score is the good outcome, so right always means the outcome you’d want. The figures on the right are as published. These are different outcomes in different people, so don’t read the rows as a ranking. None of them tested Hum.

Effect sizes for what Hum draws on, with 95% confidence intervals
What was testedWorseBetterPublished figure, 95% CI
Noise for attention, with ADHD or many ADHD traitsWhite and pink noise on lab tasks. Mostly young people. g = 0.249 (0.135 to 0.363)13 studies, N = 335Moderate. [4]
Noise for attention, without attention difficultiesSame review, comparison groups. They got worse. g = −0.212 (−0.355 to −0.069)11 comparisons, N = 335Moderate. [4]
Music and anxiety, healthy samplesSelf-reported. Physiological change wasn’t significant, and the studies vary and are often weak. d = −0.30 (−0.55 to −0.04)Study count not in our notesModerate. [5]
Natural sounds, stress and annoyanceMostly real recordings. Hum’s nature sounds are synthesised and untested. g = −0.60 (−0.97 to −0.23)18 publicationsModerate for real recordings, none for Hum’s. [6]
Breathwork, stressThe benefit is from the breathing, not from any sound. g = −0.35 (−0.55 to −0.14)12 RCTs, N = 785Moderate for breathing, none for the sound. [7]
The 16 Hz pulse in musicWoods 2024. Only a three-way interaction is reported for rate, so there’s no plain effect size to plot. Can’t be plotted F(1,80) = 4.03, p = 0.048Partial eta squared 0.048, N = 82Weak. [2] [3]
Music for insomnia, on its own scale
Different scale: points on a sleep questionnaireWorseBetterPublished figure, 95% CI
Bedtime music, adults with insomniaSelf-rated sleep quality. Objective measures, from 136 participants, may not improve. PSQI −2.79 (−3.86 to −1.72)13 RCTs, N = 1,007Moderate. [1]

g and d are standardised differences: 0.2 is usually called small and 0.5 medium. PSQI is a sleep questionnaire, where lower is better. We read many of these papers as abstracts only, marked in the sources below.

The pulse, in full

The idea comes from Woods and colleagues, 2024 [2]. In its only controlled test of rate, 82 online adults did a task for 5 minutes per condition with music pulsing at 8, 16 or 32 Hz. People who reported more ADHD traits did better over time at 16 Hz. That rests on one three-way interaction with p = 0.048. A further 93 people tested low, medium and high depth, and the paper reports no significant effects, with ratings reversing from medium to high.

The conflict. The first author is a Brain.fm employee with equity options, and another author held equity. The paper first declared no competing interests, and the journal corrected that in April 2026 [3]. The work was funded by an NSF STTR grant. We found no independent replication.

How Hum differs from that study.

A 2026 scoping review of music-based digital therapeutics reached the same general view: evidence for specific products is too limited to judge real-world effect [18]. So we rate the pulse weak. For focus it starts at Off, Low or Medium depending on your answers, and everywhere else it starts off.

Noise, reading and sleep

Noise. It can help some people and hurt others [4]. The review found no studies of brown noise, and we found none of grey. The author also warned about hearing risk with headphones and unknown long-term effects in children [19].

Reading. Across 65 studies, background speech, music and noise all had a small harmful effect on comprehension (g = −0.26, −0.19 and −0.17), and lyrical music was worst (g = −0.35) [20]. Silence is a fair choice, and Hum says so.

Sleep. Calm music helps people rate their sleep better [1]. Continuous noise is less clear: one review rated the evidence very low quality [21]. In a seven-night sleep-lab study of 25 young, healthy people, pink noise at 50 dBA cut REM sleep from 118.1 to 99.5 minutes, and earplugs did better against traffic noise [16]. So Hum keeps noise on a timer at night, and suggests earplugs first.

How strong each part is

Moderate means a meta-analysis or moderate-certainty review, usually with small effects and real limits. Weak means one or a few small, indirect, conflicted or contradictory studies. None means we found no human behavioural study of this specific thing.

Evidence strength for each part of Hum
WhatStrengthWhat Hum does
A 16 Hz pulse, for people reporting attention troubleWeakLow or Medium, depending on your task and how often you say you lose focus
Noise, with ADHD or many ADHD traitsModerate for noise alone, none combined with the pulseA little pink noise, unless you’re reading or studying, or want less sound
Noise, without attention difficultiesModerate, and it’s a small harmOff
Keeping sound low and wordless for hard or reading tasksModerateMusic lower in the mix, sparse tracks
Bedtime music for 25 to 60 minutesModerate, on self-rated sleep45 minutes, then a 15-minute fade
Continuous noise all nightWeak, with a cautionEarplugs suggested first, and always on a timer
Music for stressModeratePulse off, music you pick
Slow breathingModerate for breathing, none for the soundA swell at about 5.5 a minute to follow
The pulse for relaxing, meditating or sleepNoneOff
Binaural beats and isochronic tonesWeak for binaural, none to weak for isochronicOff, manual only
40 Hz, Schumann, SolfeggioNoneOff, manual only, no claim

What we’d never claim

What your ratings can and can’t show

A rating answers “did this feel good to you”, not “did this improve your focus”. Unusually low ratings also tend to rise on their own [22], which makes any change look like it worked. So Hum compares two setups by alternating them, over 8 rated sessions each, and keeps the new one only if it wins by 0.75 points or more.

That’s a blunt tool, and here’s how blunt. In a simulation, this rule acts on 6% of comparisons with no real difference, 28% with a real gap of half a point, and 63% with a gap of a full point. Those figures assume ratings vary by 1 point within a person, which is our assumption and not something we’ve measured. Hum can catch big differences and will rarely catch small ones.

Listening safely

Keep the volume comfortable: WHO suggests no more than 60% of your device’s maximum [23]. Hum doesn’t know how loud your device is, so it can’t promise a safe level. At night, use the lowest level you can still hear, on a speaker rather than headphones. If you have epilepsy or sound-triggered seizures, ask your doctor first. Using Hum for a child calls for lower volume, with no noise layer at bedtime [24].

Nothing on this site or in the app flashes or moves with the pulse. 16 Hz is inside the range of flashing most likely to provoke photosensitive seizures [25] [26], so we keep visuals slow and unconnected to the sound.

Set up my soundHear it first

14 days free. No card.

Sources

Numbered in the order they appear above, with how much of each we read. The app’s own suggestions draw on more papers than this page does.

  1. [1] Jespersen KV, Pando-Naude V, Koenig J, et al. Listening to music for insomnia in adults. Cochrane Database Syst Rev 2022. doi.org/10.1002/14651858.CD010459.pub3 Read the web page
  2. [2] Woods KJP, Sampaio G, James T, et al. Rapid modulation in music supports attention in listeners with attentional difficulties. Communications Biology 2024;7:1376. doi.org/10.1038/s42003-024-07026-3 Read in full
  3. [3] Author Correction (competing interests). Communications Biology 2026;9:527. doi.org/10.1038/s42003-026-10015-3 Read in full
  4. [4] Nigg JT, Bruton A, Kozlowski MB, Johnstone JM, Karalunas SL. Do white noise or pink noise help with task performance in youth with ADHD or with elevated attention problems? J Am Acad Child Adolesc Psychiatry 2024;63(8):778-788. doi.org/10.1016/j.jaac.2023.12.014 Abstract only
  5. [5] Panteleeva Y, Ceschi G, Glowinski D, Courvoisier DS, Grandjean D. Music for anxiety? Meta-analysis of anxiety reduction in non-clinical samples. Psychol Music 2017. doi.org/10.1177/0305735617712424 Abstract only
  6. [6] Buxton RT, Pearson AL, Allou C, Fristrup K, Wittemyer G. A synthesis of health benefits of natural sounds and their distribution in national parks. PNAS 2021. doi.org/10.1073/pnas.2013097118 Abstract only
  7. [7] Fincham GW, Strauss C, Montero-Marin J, Cavanagh K. Effect of breathwork on stress and mental health. Sci Rep 2023. doi.org/10.1038/s41598-022-27247-y Abstract only
  8. [8] Robison MK, Obulasetty M, Blais C, Wingert KM, Brewer GA. The effect of binaural beat stimulation on sustained attention. Psychol Res 2022. doi.org/10.1007/s00426-021-01524-3 Abstract only
  9. [9] Orozco Perez HD, Dumas G, Lehmann A. Binaural beats through the auditory pathway. eNeuro 2020. doi.org/10.1523/ENEURO.0232-19.2020 Abstract only
  10. [10] Ingendoh RM, Posny ES, Heine A. Binaural beats to entrain the brain? A systematic review. PLoS One 2023. doi.org/10.1371/journal.pone.0286023 Abstract only
  11. [11] Akkurt O, Ekici Ö. Effect of binaural beats and isochronic tones on anxiety during impacted mandibular third molar surgery. Int J Oral Maxillofac Surg 2026. doi.org/10.1016/j.ijom.2026.07.006 Abstract only
  12. [12] Chan D, Suk HJ, Jackson BL, et al. Gamma frequency sensory stimulation in mild probable Alzheimer's dementia patients. PLoS One 2022. doi.org/10.1371/journal.pone.0278412 Abstract only
  13. [13] Huang YS, Tang I, Chin WC, et al. The subjective and objective improvement of non-invasive treatment of Schumann resonance in insomnia. Nat Sci Sleep 2022. doi.org/10.2147/NSS.S346941 Abstract only
  14. [14] Bozok ÜG, Özcan GB, Bayraktar B, Özen D. Acute music-frequency exposure modulates salivary stress and neurotrophic markers in young adults. Brain Behav 2026. doi.org/10.1002/brb3.71452 Abstract only
  15. [15] Wunderlin M, Züst MA, Hertenstein E, et al. Modulating overnight memory consolidation by acoustic stimulation during slow-wave sleep. Sleep 2021. doi.org/10.1093/sleep/zsaa296 Abstract only
  16. [16] Basner M, Smith MG, Cordoza M, et al. Efficacy of pink noise and earplugs for mitigating the effects of intermittent environmental noise exposure on sleep. Sleep 2026;49(5):zsag001. doi.org/10.1093/sleep/zsag001 Read in full
  17. [17] Zavecz Z, Nagy T, Galkó A, Nemeth D, Janacsek K. The relationship between subjective sleep quality and cognitive performance in healthy young adults. Sci Rep 2020. doi.org/10.1038/s41598-020-61627-6 Abstract only
  18. [18] Venkatesan T, Demetriou AM, Hempel A, Bowling DL. A scoping review of music-based digital therapeutics for stress, anxiety, and depression. Front Hum Neurosci 2026. doi.org/10.3389/fnhum.2026.1602004 Abstract only
  19. [19] University press coverage of the Nigg review, with Nigg's comments, Medical Xpress, August 2024. medicalxpress.com/news/2024-08-white-pink-noise-focus-children.html Read the web page
  20. [20] Vasilev MR, Kirkby JA, Angele B. Auditory distraction during reading: a Bayesian meta-analysis of a continuing controversy. Perspect Psychol Sci 2018. doi.org/10.1177/1745691617747398 Read in full
  21. [21] Riedy SM, Smith MG, Rocha S, Basner M. Noise as a sleep aid: a systematic review. Sleep Med Rev 2021. doi.org/10.1016/j.smrv.2020.101385 Abstract only
  22. [22] Barnett AG, van der Pols JC, Dobson AJ. Regression to the mean: what it is and how to deal with it. Int J Epidemiol 2005. doi.org/10.1093/ije/dyh299 Abstract only
  23. [23] WHO. Deafness and hearing loss: safe listening (Q&A). www.who.int/news-room/questions-and-answers/item/deafness-and-hearing-loss-safe-listening Read the web page
  24. [24] Hugh SC, Wolter NE, Propst EJ, et al. Infant sleep machines and hazardous sound pressure levels. Pediatrics 2014. doi.org/10.1542/peds.2013-3617 Abstract only
  25. [25] Epilepsy Society. Photosensitive epilepsy. www.epilepsysociety.org.uk/about-epilepsy/epileptic-seizures/seizure-triggers/photosensitive-epilepsy Read the web page
  26. [26] W3C. Understanding Success Criterion 2.3.1: Three Flashes or Below Threshold. www.w3.org/WAI/WCAG21/Understanding/three-flashes-or-below-threshold.html Read the web page