VōxSōmaHear a free preview

By Ramūnas Deniušis, founder · September 12, 2026

How Much Deep Sleep Do You Actually Need?

You opened an app this morning and it told you that you got 48 minutes of deep sleep. Somewhere below that number there was probably a small grey bar, and somewhere in your chest there was probably a small grey feeling.

Here is the thing nobody puts next to that number: there is no official target. Not from the American Academy of Sleep Medicine, not from the Sleep Research Society, not from anyone. The target does not exist, and the number you are comparing against it is less precise than the two decimal places imply.

Part of our sound & sleep cluster.

How much deep sleep do you actually need?

There is no established recommended amount of deep sleep. The major sleep bodies publish guidance on total sleep duration — seven or more hours per night for adults — and no guidance at all on how many minutes of N3 that should contain. Deep sleep typically makes up a modest minority of the night in healthy adults, it varies enormously between people, and it declines with age in everyone.

That is the honest answer, and it is worth sitting with for a second, because almost every page that ranks for this question invents a percentage anyway.

The one real consensus statement on adult sleep need was published by a fifteen-expert panel convened jointly by the American Academy of Sleep Medicine and the Sleep Research Society (Watson et al., J Clin Sleep Med, 2015). Its recommendation is one sentence long: adults should sleep 7 or more hours per night on a regular basis to promote optimal health. The panel reviewed 5,314 papers across nine health categories to arrive at it. Sleep stages are not mentioned. There is no "and 20% of that should be N3."

What is deep sleep, exactly?

Deep sleep is stage N3 — the part of non-REM sleep dominated by large, slow, high-amplitude brain waves. It is the hardest stage to be woken from, it clusters in the first half of the night, and it is the stage associated with the slow oscillations that show up in memory-consolidation research. It is not "good sleep" and the other stages are not "bad sleep"; it is one scored category out of four.

If you want the full picture of how a night moves through those categories, we mapped it in what happens to your brain waves as you fall asleep.

What percentage of the night should be deep sleep?

No percentage has been validated as a health target. Published reference data do exist, but the researchers who produced them explicitly warn against using them as a personal benchmark — the variation between healthy people, and within the same healthy person from night to night, is large enough that a "normal range" barely constrains anything.

That warning is not our editorial spin. It is the stated conclusion of the largest modern reference-data paper in the field. Analysing 206 healthy adults aged 19 to 73 at the Freiburg sleep centre, the authors concluded that sleep parameters show "high standard deviations, suggesting a high variability within healthy adults and complicating the specification of reference values" — and then went further:

"In the absence of such knowledge, giving patients feedback about deviations from the norm in their sleep profile may give rise to ill-founded concerns and worry." — Hertenstein et al., J Clin Sleep Med, 2018

A sleep clinic is careful about showing a patient their own stage percentages, because the number causes worry out of proportion to what it means. Your wrist shows it to you every morning at 7am with a frowning face next to it.

Does deep sleep decrease with age?

Yes, and this is one of the most consistent findings in sleep research. The percentage of slow-wave sleep declines across adulthood — meaning a healthy 50-year-old comparing their number to a healthy 25-year-old's is comparing two correct answers to two different questions.

The definitive source here is a meta-analysis of 65 studies covering 3,577 people aged 5 to 102 (Ohayon et al., Sleep, 2004). In adults, it found, total sleep time, sleep efficiency, percentage of slow-wave sleep, percentage of REM sleep and REM latency all significantly decreased with age, while sleep latency, stage 1, stage 2 and wake-after-sleep-onset increased.

One detail from that paper deserves more attention than it gets: the effect sizes were "greatly modified by the quality of subject screening, diminishing or even masking age associations." In plain terms, how much deep sleep looks like it falls with age depends heavily on how carefully the studies excluded people with undiagnosed sleep disorders. Even the decline — the most robust finding available — is softer and messier than a single curve on a wellness app suggests.

Can my watch actually measure deep sleep?

Not well. Consumer wrist wearables are good at telling sleep from wake and considerably worse at sorting sleep into stages, and deep sleep is one of the stages where the minute-level errors are largest relative to the quantity being measured.

A 2025 validation study put six commercial wrist-worn devices against full polysomnography on 62 adults in a hospital sleep laboratory (Schyvens et al., SLEEP Advances, 2025). Every device detected over 90% of sleep epochs — they know when you are asleep. But specificity ranged from 29.4% to 52.2%, and agreement with polysomnography across stages produced Cohen's kappa values of 0.21 to 0.53: fair to moderate. The authors' own conclusion is that "all devices can benefit from further improvement for multistate categorization."

Here is what that meant for deep sleep specifically. Device names are omitted deliberately — the point is the category, not a brand comparison.

Device Nights analysed Lab-measured deep sleep Device's estimate Error
A 37 68.8 min 65.0 min −3.9 min (not significant)
B 39 65.5 min 63.3 min −2.2 min (not significant)
C 40 60.6 min 92.1 min +31.5 min
D 25 75.9 min 120.4 min +44.4 min
E 20 71.0 min 45.8 min −25.2 min

A sixth device was excluded from this comparison because it reports deep and REM sleep as one combined figure. Bold = statistically significant difference from polysomnography.

Read the two "not significant" rows carefully, because they are the most misleading ones. Devices A and B were accurate on average across the group — but their mean absolute error per night was around 26 minutes each. Against a true value of roughly 65–69 minutes, that is an error of about 40% on any given night. The group average is right; your Tuesday is a coin flip.

So when a number drops from 71 minutes to 45 minutes overnight and you spend the day feeling like you have damaged something, there is a real chance the only thing that changed was the estimate.

Is worrying about your deep sleep score making things worse?

It can. Clinicians have a name for it. In 2017 a team described a pattern of patients arriving at sleep clinics with self-diagnosed sleep problems based on tracker data, and coined orthosomnia — from ortho, "correct" — for the perfectionistic pursuit of ideal sleep (Baron et al., J Clin Sleep Med, 2017).

The paper's most uncomfortable observation is that for these patients, the tracker's data "often feels more consistent with their experience of sleep than validated techniques, such as polysomnography." The number wins the argument against the laboratory. It will certainly win the argument against your own morning.

This is the quiet trap in the sleepmaxxing trend: the measuring becomes the ritual, and the ritual becomes the thing keeping you up.

What should you pay attention to instead?

Total sleep duration, regularity, and how you feel during the day — all three are better supported by evidence, and unlike stage percentages, all three are things you can actually act on.

What you're tracking Evidence behind a target Can you change it directly?
Deep sleep minutes No published target No — not a voluntary behaviour
Total sleep duration Strong: ≥7 hours, expert consensus Yes — bedtime and wake time
Sleep regularity Good and growing Yes — same schedule, including weekends
Time to fall asleep Reasonable Partly — via wind-down conditions
Daytime alertness The outcome that actually matters Indirectly, via the above

Deep sleep is not a behaviour. There is no lever for it. It is a downstream result of sleeping enough, at consistent times, without something interrupting you — which is why the honest advice for "getting more deep sleep" is identical to the advice for getting more sleep of every kind: protect the hours, keep the schedule, and make the last thirty minutes before bed genuinely dull.

If that last part is where you struggle, the practical pieces are building an evening wind-down routine, slow breathing before sleep, and — if the real obstacle is that you keep postponing bed for the only quiet hour you get — revenge bedtime procrastination.

Where VōxSōma sits in this

VōxSōma measures nothing. There is no sensor, no score, no morning number, and no stage graph. It cannot tell you how much deep sleep you got and does not try, because the honest version of this product is a wind-down ritual, not a measurement device — and because, as the research above makes clear, a wrist-based stage number is not a thing worth building a product around.

What it is: a five-layer audio architecture built around seven affirmations recorded in your own voice, arranged as a 36-minute structured descent. One purchase, no subscription, and the voice never leaves your device. You can hear how it is built, listen to a preview, read why it exists, or see what it costs.

It will not raise your deep-sleep score. Nothing you buy reliably will, and anything claiming otherwise is selling you a target that no sleep organization has ever published.

FAQ

Is 45 minutes of deep sleep enough? There is no threshold that makes a number "enough." No sleep organization publishes a minimum for N3, published reference ranges vary widely between healthy adults, and a wrist-measured figure can be off by tens of minutes on any individual night. Forty-five minutes on a watch is not evidence of a problem.

How can I increase my deep sleep? There is no reliable, direct method. Deep sleep is not a voluntary behaviour — it is a consequence of sleeping enough, on a consistent schedule, without disruption. Protecting total sleep duration and keeping regular hours is the whole of the evidence-backed answer.

Do you get less deep sleep as you get older? Yes. The percentage of slow-wave sleep declines significantly across adulthood, confirmed by meta-analysis across 65 studies and 3,577 people. It is a normal feature of ageing, not a fault to correct, and it means age-matched comparison is the only comparison that means anything.

Are sleep trackers accurate for sleep stages? They are good at sleep-versus-wake and fair-to-moderate at staging. In a 2025 lab validation of six wrist wearables, devices detected over 90% of sleep epochs but agreement with polysomnography ranged from kappa 0.21 to 0.53, and several devices mis-estimated deep sleep by 25 to 44 minutes on average.

Can audio, binaural beats or pink noise give me more deep sleep? The evidence does not support a promise. Binaural beats remain mixed evidence at best. Pink noise is the stimulus used in closed-loop acoustic stimulation research, but that technique requires live EEG measurement and even there the behavioural benefits do not replicate consistently. Ordinary playback of pink noise from a speaker is not the same intervention.

Should I stop tracking my sleep entirely? Not necessarily — trackers are genuinely useful for duration and regularity, which are the two things worth acting on. The suggestion is narrower: stop treating the stage breakdown as a grade. If checking the number is itself making your nights worse, that is a good reason to turn that screen off.

Sources

VōxSōma is a personal wellness audio tool — not a medical device, not therapy, and not intended to diagnose, treat, cure, or prevent any condition. Individual experiences vary. If you have a sleep, attention, or mental-health condition, please speak with a qualified clinician.