Thomas Stephan Juzek

LIN 3042 Applied Linguistics · Fall 2026

Sleep sounds

From a question in class: brain activity in sleep can be drawn as waves. Is that like sound, and can you hear it? Yes, if you speed it up. Not on the quiz.

AI disclaimer. Content guided by Tommie Juzek; built upon his prompting by an AI assistant (Claude Opus 5.5, Anthropic), which also chose the clips and checked the sources; reviewed and approved by Tommie Juzek.

Are brain waves like sound waves?

In part. The waves in a sleep recording are EEG (electroencephalography): electrodes on the scalp pick up tiny changes in voltage, millionths of a volt, produced when large numbers of brain cells are active together. It is the same recording that ERPs are made from (textbook pp. 433–434). An ERP averages many short stretches of EEG, each lined up with a word; a sleep recording keeps the whole night and is read by its rhythms.

What EEG shares with sound: both are a single value changing over time, drawn as a wave, and both can be taken apart into frequencies. Sleep researchers read frequency bands much as phoneticians read a spectrogram.

What differs: sound is air pressure; EEG is voltage, and nothing vibrates. And the speed. The main sleep rhythms run at about 0.5 to 16 cycles per second (hertz, Hz); human hearing starts at about 20 Hz. Played as it is, a night of EEG is silent.

So this page plays it 200 times faster. A 1 Hz wave becomes 200 Hz, a low hum, and 30 seconds of sleep pass in 0.15 seconds. This is called audification. It is not new: the researcher who shared these recordings published tools for listening to sleep recordings this way (Olivan, Kemp, & Roessen, 2004). And EEG turned into sound has real uses: in one study, people with no EEG training picked out seizures by ear, from EEG that steered a tone (Parvizi et al., 2018). One caveat: you hear the rhythm of the brain, not its thoughts. There are no words in here.

Speed

Play faster than real time

Listening: room speakers or headphones work best. Laptop and phone speakers lose the deep rumble, especially at 100×. On an iPhone, if you hear nothing, flip the ring/silent switch to ring.

Five stages of sleep

Each picture is one 30-second page of the recording, the unit sleep technicians score. Every clip uses the same volume setting, so a louder clip means bigger waves. Download saves a clip once, as a WAV file, at the speed set above.

Awake, relaxed W

Listen for: a high tone over the hum. That is the alpha rhythm, about 10 Hz (2 kHz at 200×), strongest when the eyes are closed. It shows best at the back of the head, so this clip, unlike the others, comes from the electrode pair there (Pz-Oz).

Light sleep N1

Listen for: small, mixed waves, slower than alpha (theta, 4–7 Hz). N1 is the doorway into sleep and rarely lasts long: this was the longest stretch of the night, 3 minutes, so the clip loops.

Sleep stage 2 N2

Listen for: a faint, high crackle on top of the hum. These are sleep spindles, short bursts of 11–16 Hz waves, about four a minute in this clip. Sped up, each burst lasts only a few thousandths of a second, so you hear a crackle rather than separate chirps; try 100×. In the picture they are the short bursts of fast, dense waves.

Deep sleep N3

Listen for: a loud, low rumble. Slow waves, 0.5–2 Hz (100–400 Hz at 200×), are the biggest waves of the night. This clip is louder because they really are bigger.

REM sleep REM

Listen for: the quietest clip: small, mixed waves, with no spindles and no slow waves. In REM (rapid eye movement) sleep the eyes dart about and the body’s muscles go slack. Through this one brain channel, REM sounds much like light sleep; technicians tell the two apart mainly with the eye and chin channels, which are not played here.

A whole night in about two minutes

Seven hours of one person’s night, from half an hour before falling asleep to half an hour after waking. The chart is the hypnogram, the technicians’ stage for each 30-second page. Tap the chart to jump.

Look and listen for: deep sleep (N3, the loud rumble) comes mostly early: 87 of this night’s 110 minutes of N3 fall in the first half. Sleep runs in cycles: REM (thick line) comes back four times, about every 70 to 75 minutes in this night.

Guess together

Three clips from a different sleeper. Which stage is each? There are five to choose from (W, N1, N2, N3, REM), so guessing alone gets one in five right.

Clip A

Clip B

Clip C

The recordings and the sources

Two nights from the Sleep-EDF Database Expanded on PhysioNet (recordings SC4001 and SC4011), from a study of sleep in healthy adults aged 25 to 101 who took no sleep medication. Trained technicians scored each 30-second page. They used the older system with stages 1 to 4; here stages 3 and 4 are combined as N3, as in current practice. The recordings are shared under the Open Data Commons Attribution License v1.0; unlike the rest of this site, they are not CC0. The downloads fall under the same licence: if you share them, credit the Sleep-EDF Database Expanded on PhysioNet.

What was done to them: filtered to 0.3–30 Hz, and one volume setting for every clip. The clips were chosen by rule, not by ear: for each stage, the longest unbroken stretch of the night (its middle 10 minutes, if longer); for the awake clip, the 10 minutes after the final awakening. Nothing is stored or sent from this page.

  1. Kemp, B., Zwinderman, A. H., Tuk, B., Kamphuisen, H. A. C., & Oberyé, J. J. L. (2000). Analysis of a sleep-dependent neuronal feedback loop: The slow-wave microcontinuity of the EEG. IEEE Transactions on Biomedical Engineering, 47(9), 1185–1194. The publication PhysioNet asks users of these recordings to cite.
  2. Pollard, T., Moody, B. E., Lehman, L., Gow, B., Fernandes, C., Xie, C., Johnson, A., Mark, R. G., & Heldt, T. (2026). PhysioNet as a global platform for biomedical research. Nature Health, 1(8), 792–795.
  3. Olivan, J., Kemp, B., & Roessen, M. (2004). Easy listening to sleep recordings: Tools and examples. Sleep Medicine, 5(6), 601–603.
  4. Parvizi, J., Gururangan, K., Razavi, B., & Chafe, C. (2018). Detecting silent seizures by their sound. Epilepsia, 59(4), 877–884.
  5. O’Grady, W., Archibald, J., Aronoff, M., & Rees-Miller, J. (Eds.). Contemporary Linguistics (7th ed.), pp. 433–434: ERPs and EEG.

LIN 3042 course page