Speech in Noise Test

By Murat Şahin · Updated November 12, 2026 · How we test

What this test measures

Hearing a quiet beep and following a voice in a noisy room are two different skills. A hearing test with tones checks the first. This speech in noise test checks the second: how much louder than the background a voice has to be before you understand it.

The method is the digits-in-noise test, developed by Cas Smits and colleagues in the Netherlands as a telephone hearing screen (2004) and used in the World Health Organization's hearWHO app. Because it compares voice with noise, the result depends far less on your volume setting than a tone test does.

How the test works

Pick a voice and an ear setup. Female is the default voice. Same in both ears is the standard setup; Antiphasic phase-inverts the voice in one ear while the noise stays identical, and needs stereo headphones.
Set a comfortable volume. Digits play over a soft hiss, fading in from silence. Make the voice clear and comfortable, then press The volume is set. Voice plus noise stays at this total loudness for the whole test, so nothing gets louder than what you chose.
24 rounds of three digits. Three different digits follow each other about half a second apart. Type them on the keypad or keyboard and press OK. A round is correct only if all three are right and in order; if unsure, guess.
The noise adapts. The first round plays at 0 dB SNR (voice and noise equally loud). After a correct round the voice gets 2 dB quieter relative to the noise; after a wrong round, 2 dB louder. This one-up, one-down rule settles where you get about half of the rounds right. As in the original test, there is no feedback along the way.

A run takes about three minutes. The digits are short recordings stored on this site; nothing is sent anywhere.

How to read your result

Your score is the speech reception threshold (SRT): the average signal-to-noise ratio of rounds 5 to 24, in dB SNR. Rounds 1 to 4 only home in on your level. Lower is better: −10 dB SNR means you followed half of the triplets while the voice was 10 dB quieter than the noise.

The card also shows Rounds correct and your SNR track: filled dots are correct rounds, rings wrong ones, the dashed line your threshold. A good track zigzags around one level; one that keeps climbing usually means a distraction or a headphone problem.

For a reference point, a South African English smartphone version of the test used −9.55 dB SNR as its pass/refer cut-off for native speakers and confident non-native speakers (Potgieter, Swanepoel & Smits, 2018). The card marks it for the same-in-both-ears setup. Treat it as a pointer only: our recordings, noise and level differ from that study, so this is a non-normed adaptation, and a result near the line can land on either side another day.

Antiphasic results are not compared with that cut-off. In De Sousa and colleagues' study (2020), normal-hearing listeners scored 6 to 8 dB better antiphasically, and this version detected hearing loss above 25 dB HL more accurately. Each setup and voice keeps its own personal best.

Why understanding speech in noise is a separate skill

Picking one voice out of many is called the cocktail party effect, after Colin Cherry's experiments in 1953. The brain uses timing and loudness differences between your ears, the pitch of the voice and your knowledge of language to separate speech from noise.

That is why some people pass a tone test yet struggle in restaurants. In mice, Kujawa and Liberman (2009) showed that noise exposure can destroy many nerve connections in the inner ear even after hearing thresholds have returned to normal. The human version of this idea, hidden hearing loss, is still being studied; a tone test cannot reveal it, and speech-in-noise measures are one way researchers look for it.

What affects your result

Language. The test uses digits in the page's language. In the South African study, non-native listeners who rated their English as weak scored clearly worse than the others (Potgieter et al., 2018), so take the test in your strongest language.
Practice. New listeners need one training run before scores settle (Smits, Goverts & Festen, 2013), so treat your first run as practice.
Headphones and room. Room noise adds to the test noise; loose earbuds and Bluetooth processing change the sound.
Attention. A lapse costs 2 dB, so pick a moment when you can focus. To check your hearing at the high end, try the hearing age test; if you hear a constant ringing, the tinnitus test helps you describe its pitch.

Limits and when to see an audiologist

This is a self-check, not a diagnosis. It does not test each ear separately or tell types of hearing loss apart. See an audiologist if your threshold stays above the cut-off on a second run with good headphones in a quiet room, or if conversations in noisy places are often hard to follow. Sudden hearing loss, new ringing in one ear or dizziness need a doctor the same day. To measure how fast you react to sound instead, try the audio reaction time test.

More tests for your ears are on the hearing tests page.

Sources

Smits, C., Kapteyn, T. S., & Houtgast, T. (2004). Development and validation of an automatic speech-in-noise screening test by telephone. International Journal of Audiology, 43(1), 15–28.
Smits, C., Goverts, S. T., & Festen, J. M. (2013). The digits-in-noise test: Assessing auditory speech recognition abilities in noise. Journal of the Acoustical Society of America, 133(3), 1693–1706.
Potgieter, J.-M., Swanepoel, D. W., & Smits, C. (2018). Evaluating a smartphone digits-in-noise test as part of the audiometric test battery. South African Journal of Communication Disorders, 65(1), a574.
Potgieter, J.-M., Swanepoel, D. W., Myburgh, H. C., & Smits, C. (2018). The South African English digits-in-noise hearing test: Effect of age, hearing loss, and speaking competence. Ear and Hearing, 39(4), 656–663.
De Sousa, K. C., Swanepoel, D. W., Moore, D. R., Myburgh, H. C., & Smits, C. (2020). Improving sensitivity of the digits-in-noise test using antiphasic stimuli. Ear and Hearing, 41(2), 442–450.
Cherry, E. C. (1953). Some experiments on the recognition of speech, with one and with two ears. Journal of the Acoustical Society of America, 25(5), 975–979.
Kujawa, S. G., & Liberman, M. C. (2009). Adding insult to injury: Cochlear nerve degeneration after "temporary" noise-induced hearing loss. Journal of Neuroscience, 29(45), 14077–14085.
World Health Organization (2019). hearWHO: hearing screening app.

Frequently asked questions

What is a speech in noise test?

It measures how well you understand speech when there is background noise. In this version you hear three spoken digits in a steady hiss and type them in. The noise adapts to your answers until it finds the signal-to-noise ratio at which you get about half of the triplets right.

What is a good digits-in-noise score?

Lower is better. A South African English smartphone version of the test used -9.55 dB SNR as its pass/refer cut-off for native and confident non-native speakers. Our recordings and noise are different and not normed, so use that value only as a rough pointer and compare your own runs with each other.

Why does the test use digits instead of words or sentences?

Digits are known to almost everyone who speaks a language, including children and people with a small vocabulary, so the test measures hearing in noise rather than vocabulary. That is one reason digits are used in telephone and smartphone hearing screens.

Do I need headphones for the speech in noise test?

Yes. Headphones keep room sound and echo out and send the same signal to both ears. The antiphasic setup only works with stereo headphones; on a speaker or a mono output the inverted voice can cancel out.

What does antiphasic mean?

In the antiphasic setup the voice is phase-inverted in one ear while the noise is identical in both ears. The brain can use that difference between the ears to pull the voice out of the noise, so results are usually several dB better than with the same signal in both ears. Each setup keeps its own personal best.

I passed a tone hearing test but struggle in noisy places. Why?

Understanding speech in noise depends on more than hearing quiet tones. The brain combines information from both ears, the voice and your knowledge of language. Research on so-called hidden hearing loss suggests some damage does not show up on a tone test. If noisy places are often hard for you, an audiologist can test speech in noise properly.

Is this test a diagnosis?

No. It is a self-check with uncalibrated headphones and our own recordings. It does not test each ear separately and cannot tell types of hearing loss apart. See an audiologist if your result stays above the cut-off on a second run, and a doctor the same day for sudden hearing loss, new ringing in one ear or dizziness.