The best sound to wake up to

The research on which sounds wake a sleeping person was not done by sleep scientists. It was done by fire-safety researchers, because for them it is a question about people dying rather than a question about preference. Their consistent finding, across every population tested: the standard high-pitched beep is one of the worst waking signals available. In a study of 176 children, a low-pitched tone of roughly 500 hertz woke 88%, a recorded voice woke about 83%, and the conventional high-frequency alarm woke 49% (Smith 2020).

What the fire research found

The pattern has now been replicated across quite different groups. In 38 adults aged 18 to 80 with mild to moderately severe hearing loss, a 520 hertz square wave woke 92% at the 75 dBA bedroom standard where the conventional 3100 hertz alarm woke 56%, and strobe lights woke 27% even when run above the US standard (Bruck & Thomas 2009). In 39 young adults tested across three nights, the lowest waking thresholds belonged to the low-frequency square waves at 400 and 520 hertz, beating pure tones, whoops, and white noise (Bruck 2009). A review of the field reached the same conclusion about voice alarms and mixed-frequency signals against the high beep (Bruck & Ball 2007).

The voice results are the surprising part. When 24 children were exposed to both their own mother’s recorded voice and a conventional tone, on separate cycles of deep sleep, the voice woke 96% against the tone’s 58%, and the median time to wake was 20 seconds against three minutes (Smith 2006). The obvious explanation, that a familiar and loved voice cuts through, turns out to be wrong. In the four-signal study, a female stranger’s voice woke 83% and the child’s own mother’s voice woke 78%, a gap that did not reach significance, and a separate trial found that including the child’s own first name added nothing (Smith 2019). Speech itself seems to be what carries, not who is speaking.

Why you should not simply switch your phone to a 500 hertz tone

Four differences between that research and your morning, and they all cut the same way.

Volume. The child studies ran at 85 to 100 decibels at the pillow, and the hearing-loss study at 75. A phone on a nightstand is well below any of that. The one study of children run at a quiet 60 decibels found that 85% of those aged 6 to 17 slept through at least one presentation of a high-pitched alarm, which is at least consistent with level mattering, though that study compared children against adults rather than loud against quiet (Bruck 1999).

Sleep stage. Every one of these trials deliberately fired the alarm during the deepest stage of sleep, because that is the hardest state to wake from and the one that kills people in fires. A 7am alarm arrives when deep sleep is largely spent. These are worst-case numbers by design.

Novelty. A smoke alarm is a sound you have essentially never heard. Your morning alarm is one you have heard several hundred times, and habituation is the single largest reason not to move a number from one setting to the other. The researchers themselves noted that children in one study were primed and motivated to beat the alarm.

Who has the problem. In a later study by the same group of 150 healthy adults aged 20 to 49 (Smith 2020b), every signal woke essentially everyone, and the difference between the best and worst was a matter of seconds rather than of waking at all. If you are a healthy adult, the signal type is probably not what is wrong.

What about grogginess?

A different question from waking, and the one most people actually care about, since being awake at 7am is easy and being functional at 7:15 is not. That fogginess has a name, sleep inertia, and it is driven mostly by which stage of sleep you were pulled out of rather than by what pulled you.

On the sound specifically, the evidence is two small studies from the same group. A survey of 50 people found no significant association between grogginess and the type of waking sound, but did find that sounds people rated as melodic were linked to reports of less grogginess, while sounds rated as neither melodic nor unmelodic were linked to reports of more (McFarlane 2020). A follow-up with 20 people at home found a melodic treatment reduced attention lapses on a reaction-time test, though it did not change reaction time itself (McFarlane 2020b).

Twenty people split into two groups is a very thin basis for anything, and the effect is on a laboratory attention measure rather than on how your morning goes. The same team’s systematic review is blunt about the gap: musical approaches have not been tested in emergencies, and alarm and voice signals have not been tested outside them.

So what should you set?

Given all of the above, this is a section of reasonable inferences rather than findings, and it is worth saying so.

• Choose something low and tonal over something high and piercing. The direction of the evidence is consistent across every population tested, even if the volumes were not yours.
• Prefer a melody to a bare beep, on the weak evidence above, and because nothing suggests the beep has any advantage.
• Change it when it stops working. Habituation is the mechanism the research cannot measure and the one you can observe directly. A sound that has stopped waking you has told you something.
• Turn the volume up before you blame the sound. The one quiet-volume data point on children is consistent with level mattering, even though it was not designed to test that.
• Skip sleep-cycle alarm apps. Their premise rests on consumer trackers identifying sleep stages accurately, and how accurate sleep trackers really are is not a flattering read on that point.

The bigger lever is not the alarm at all. Waking at a consistent time, and getting light into your eyes soon after, does more for how mornings feel than any tone, because both anchor the body clock that decides how hard waking is in the first place. And the fire-safety literature has one piece of practical advice that has nothing to do with your phone: if there are children in the house, the alarm that is supposed to wake them at 3am is worth checking, because the standard one is not good at it.

---

Sources: Smith et al. 2020 · Smith et al. 2020b · Bruck 1999 · Smith et al. 2006 · Smith et al. 2019 · Bruck & Thomas 2009 · Bruck et al. 2009 · Bruck & Ball 2007 · McFarlane et al. 2020 · McFarlane et al. 2020 · McFarlane et al. 2020, systematic review. The smoke-alarm studies describe fire-safety signals at high volume and are not safety advice. This is general information, not medical advice — persistent insomnia is worth taking to a doctor.