How Accurate Are Snoring Apps? What the Studies Actually Show

What snoring apps actually measure

Every snoring app starts with the same raw material: sound picked up by your phone's microphone. From that, apps typically report some combination of recording time, how much of the night contained snoring, how often snoring occurred, how loud it was, and a proprietary score. These are separate quantities. A night can have a lot of quiet snoring, or a few very loud snores — and a single score blends them in a way that differs from app to app.

It helps to separate three jobs an app performs:

• Recording — capturing sound and, sometimes, saving clips. This is the easy part.
• Classification — deciding whether a sound is a snore, speech, breathing, a cough, traffic or a pet. This is where most validation research focuses.
• Loudness estimation — turning the microphone signal into a level. Without calibration, any absolute decibel value is an estimate.

Detecting sound is not the same as identifying snoring

A microphone hears everything. Snoring is low-frequency and rhythmic, but so are some fans and heating systems. Speech, sleep-talking and heavy breathing overlap with snoring in frequency. A partner's snoring, a dog on the bed or a television in the next room can all be counted. An app's detection algorithm has to make a judgement for each sound, and it can be wrong in both directions: counting non-snores (false positives) and missing quiet or irregular snores (false negatives).

That is why validation studies compare the app with a reference — a polygraphy or polysomnography snore sensor, or a trained person listening back. Even then, the reference itself has limits, and results depend on who was studied and where.

What the validation studies show

The research base is small. The two most cited studies of consumer snoring apps tested 19 and 11 people respectively. A larger 2024 study tested three apps against in-lab polysomnography. A separate NIOSH study tested sound-level-meter apps — not snoring apps — in a laboratory.

How to read these results

• High accuracy for heavy snoring is not the same as high accuracy for everyone. In the SnoreLab study, accuracy figures were reported for people who snored more than half of the night; the positive predictive value at that threshold was 66.6%.
• The reference matters. Snore Clock was compared with an ENT specialist listening to the app's own recordings, not with polysomnography.
• Correlation is not agreement. The 2024 polysomnography study relied mostly on correlations; its authors note this does not show that every event was captured.
• None of these findings transfer to other apps — including SnapSnoreScore.

Absolute decibels vs. change above background

Many apps display a number like "62 dB". For that number to be an absolute sound pressure level, the phone's microphone chain would need to be calibrated against a reference meter. Phones are not shipped that way, and models differ in sensitivity, frequency response and built-in audio processing. In a NIOSH laboratory evaluation, most sound-meter apps screened did not meet the study's criteria; a handful of apps on specific devices came within about 2 dB of the reference.

A more robust measure is how far snoring rises above the room's own background, recorded by the same phone in the same place. Errors in the microphone's absolute sensitivity largely cancel out when you compare two levels from the same device. That is why a rise of "+18 dB above background" is more meaningful for night-to-night comparison than an uncalibrated "62 dB".

Our separate guide, How to Measure Snoring Decibels at Home, walks through setting this up. For typical published snoring levels and the full decibel chart, see How Many Decibels Is Snoring?

Phone, position and room change the result

The same snore can produce different readings depending on:

• The phone — microphone placement, sensitivity and any noise suppression the operating system applies.
• Distance and orientation — sound level falls with distance, and a microphone facing away or lying face-down picks up less.
• Obstruction — a phone under a pillow, in a thick case or buried in bedding is muffled.
• The room — hard walls reflect sound; curtains, carpets and bedding absorb it.

In the SnoreLab study, researchers controlled for this by asking participants to place the phone one to two metres from the pillow with the microphone towards the head, keep windows closed, and sleep without a bed partner or pet. Real homes are rarely that tidy.

Time, frequency and loudness are different things

• Recording time — how long the app ran. Not the same as sleep time.
• Snoring duration / share — how much of that time contained detected snoring.
• Snoring frequency — how many snore events per hour.
• Loudness — how loud the snores were, as a peak, an average or a rise above background.

Two nights with the same score can look very different on these measures. When comparing nights, look at the parts, not just the total.

Why snoring scores differ between apps

Snoring scores are proprietary. The SnoreLab study authors pointed out that its Snore Score is "based on an unknown algorithm" and cannot be directly compared with clinical scores. The same applies to every app's score, including ours. A score is a summary chosen by the app's designers — useful for tracking yourself on one app, not for comparing yourself with someone using a different app.

Why an app can't diagnose or rule out sleep apnea

Obstructive sleep apnea is defined by repeated narrowing or closure of the airway during sleep, measured as breathing pauses and reductions in airflow, usually with oxygen levels. A sleep study measures these directly. Snoring audio does not.

• In the SnoreLab study, obstructive events could not be detected by the app.
• In the 2024 study, some app scores correlated with apnea severity across the group — but a group-level correlation cannot tell an individual whether they have sleep apnea.
• People with sleep apnea do not always snore loudly, and many loud snorers do not have sleep apnea.

So a quiet result does not rule sleep apnea out, and a loud result does not confirm it. Read more in Snoring vs. Sleep Apnea.

Where long-term tracking can help

With the same phone, the same position and similar room conditions, an app can reasonably show direction of change: whether snoring time or loudness went down after you started sleeping on your side, avoided alcohol in the evening, or treated nasal congestion. Several nights before and after a change say more than any single night. Saved clips can also help a clinician or partner hear what the night sounded like.

When to see a doctor — whatever the app says

Speak to a doctor if you or a partner notice any of the following, regardless of any app reading:

• Pauses in breathing, gasping or choking during sleep
• Excessive daytime sleepiness, or falling asleep while driving
• Morning headaches, or waking unrefreshed despite enough time in bed
• High blood pressure alongside loud, habitual snoring
• Snoring in a child that is frequent or accompanied by breathing pauses

What SnapSnoreScore does — and doesn't

SnapSnoreScore runs in your browser. During a recording it:

1. Learns your room's background level before and during the night.
2. Applies A-weighting and analyses the sound in short frames as it arrives.
3. Uses an adaptive detector to decide which sounds are probably snores, using their frequency content, duration and rhythm.
4. Reports how far snoring rose above your room's background as the main loudness figure. An absolute dB value is shown only as an estimated range, based on the distance and room type you select.
5. Calculates a 0–100 Snore Score from four parts: share of the recording with snoring (up to 40 points), events per hour (up to 25), loudness above background (up to 25) and how regular the pattern is (up to 10).

What it is not: SnapSnoreScore has not been clinically validated, and the studies above did not test it. The Snore Score is a proprietary descriptive index, not a medical score. A phone-based estimate is not interchangeable with a calibrated sound level meter, and the app cannot reliably tell whose snoring it hears.

Privacy: audio is analysed on your device and is never uploaded. Up to 12 short snore clips per recording (about 6 seconds each) may be stored in your browser for playback, until you delete them. Details are in the Privacy Policy and FAQ.