Speech in Noise Test: Free Online Hearing Screening
Overview
Speech in Noise Test speaks three digits through your browser's own voice against a bed of background noise, then adapts the noise level based on whether you got them right — converging on the signal-to-noise ratio where you can only follow along about half the time, the same digits-in-noise approach behind large-scale hearing screening tools like the WHO's own app. Use the free hearing test whenever you want a fast, repeatable way to confirm this.
If you can hear people talking in a busy room but can't make out what they're saying, a speech in noise test puts a number on that problem in a few minutes. You listen to spoken numbers through headphones while background noise rises and falls, and the test finds the point where you catch about half of them, your SNR-50 score in decibels. A standard pure-tone test only checks your auditory thresholds in silence, so it can come back normal while you still struggle in noisy environments. This free digits-in-noise check fills that gap: set a comfortable volume, keep it fixed, and let the result show how your hearing copes with real-world clutter.
What a Speech in Noise Test Measures
A speech-in-noise test measures how well you understand speech when other sound competes with it. Instead of asking whether you can detect a sound, it asks whether you can follow the words. That's the listening skill that matters in a busy café or a family dinner, and it depends on your ear and your brain working together, not on hearing alone. It's also the part of communication most people notice first when it starts to slip.
The idea is old. Researchers argued in the 1940s that a speech-in-noise measure would capture everyday communication better than lists of words read in silence, and audiology has recommended speech-in-noise testing ever since. Even so, surveys suggest that well under half of hearing clinicians run one routinely, mostly because of test time, cost and reimbursement, so routine audiometry still often skips it. This page gives you a free online hearing screening version you can take at home, so you can see whether your hearing difficulty is about volume or about comprehension when there's noise around you, and decide what conversation you need to have next.
Speech Understanding vs Hearing Quiet Tones
The usual audiogram plots your pure-tone thresholds: the softest beep you can hear at each pitch, measured in a silent booth. It's an excellent map of sensitivity, but it says little about speech understanding in a crowd. Two people with the same audiogram can score very differently once noise is added.
A standard hearing test also often includes speech in quiet, such as a speech discrimination score, where you repeat words read at a comfortable level. Those results are useful, but SIN tests ask a harder question: how much louder than the noise does speech need to be before you follow it? That's why audiologists treat speech-in-noise testing as a separate part of the picture rather than a repeat of the tone test: it measures listening in background noise, not detection in silence.
Why Background Noise Is the Real Complaint
Ask people why they think they need hearing help and the answer is rarely "beeps sound quiet." It's "I can't follow what people say in noisy restaurants." Background noise is the number one complaint among people treated for hearing loss, and it's the setting a speech-in-noise score is built to describe. Hearing aid wearers say the same thing: quiet rooms are fine, noisy environments are where things fall apart.
Real-world rooms are rarely kind. Measurements across homes, classrooms, shops and transport put the gap between talker and noise mostly somewhere between 0 and 15 dB, and busy social settings like cocktail parties sit at the low end of that range. The listening environment also changes from minute to minute: a coffee grinder, a passing bus, two tables of competing sounds. With aging, the brain's ability to pull one voice out of that background noise tends to drop even before the audiogram shows much change, which is why listening in noisy environments often gets harder in your fifties and sixties.
How the Digits-in-Noise Test Works Online
This page uses a digits-in-noise format, the same style national hearing screening programs use, because numbers are easy to recognize in any accent and don't depend on vocabulary. Each trial plays three digits spoken over a noise background. You type what you heard, and the next trial gets harder or easier depending on your answer. The binaural beats generator online is free to use, with no limit on how long you run it.
- Put on your headphones and play the volume check.
- Set a comfortable level, then leave the volume alone for the whole test.
- Listen to each set of three digits in noise and enter them, guessing when you're unsure.
- After about 20 to 25 trials, the test reports the signal-to-noise level where you get the digits right about 50% of the time.
Keep any background noise in your own room to a minimum while you test. Like other speech-in-noise tests built for screening, it's a screening tool: quick, repeatable and good at spotting a problem, but not a diagnosis.
Setting Up Headphones, Volume and a Quiet Room
Your setup affects your score as much as your hearing does, so it's worth two minutes of care:
- Headphones: wired or Bluetooth over-ear headphones give the most consistent results. Earbuds work if they seal well; laptop speakers don't, because room echo adds its own noise.
- Quiet room: turn off fans and the TV and close the door. Any sound in the room stacks on top of the test noise and makes your score look worse than it is.
- Volume: start low and raise it to a comfortable level during the sample. Don't adjust it mid-test, because the method relies on a steady speech level.
- Browser: current versions of Chrome, Firefox, Safari and Edge work. The page plays recorded audio and doesn't depend on the Web Speech API, so voices sound the same on every device.
The Adaptive Protocol Behind Each Trial
The test follows an adaptive protocol, much like the way an audiologist brackets a tone threshold. The speech stays at a fixed, comfortable level, close to your most comfortable loudness, and the noise moves. After a correct answer the noise gets louder, so the next trial is harder. After a miss it gets quieter.
Early trials use larger step sizes so the test homes in on your range quickly. Once your answers start to flip between right and wrong, it switches to smaller 1-dB steps, which matter more than they sound: a 1 dB change in signal-to-noise ratio moves word recognition by roughly 8 to 10 percentage points near the middle of the curve. The final score averages the levels from those last reversals, which keeps it stable even if you guess on a few trials.
The SNR-50 Formula Behind Your Speech-in-Noise Score
Your result is reported as an SNR-50: the signal-to-noise ratio, in decibels, at which you understand half of what you hear. A lower number is better. A negative score means you can follow speech even when the noise is louder than the talker. The left and right ear comparison online gives you a clear answer instead of guessing.
Signal-to-Noise Ratio in dB SNR
The signal-to-noise ratio is simply the speech level minus the noise level, both in dB:
$$\text{SNR (dB)} = L_{\text{speech}} - L_{\text{noise}}$$So speech at 72 dB over noise at 66 dB is +6 dB SNR, and speech at 60 dB over noise at 63 dB is −3 dB SNR. Because decibels are logarithmic, every 6 dB of SNR means the speech carries about twice the sound pressure of the noise. That's why a few dB of difference between two listeners is a big real-world gap rather than a rounding error.
From SNR50 to SNR Loss
The raw SNR50 depends heavily on the test: digits in steady noise give lower scores than sentences in babble, so you can't compare raw numbers across different tests. The fix is to subtract the average score of young listeners with normal hearing on the same test. The difference is your SNR loss:
$$\text{SNR loss} = \text{SNR50}_{\text{you}} - \text{SNR50}_{\text{normal}}$$In the speech-in-noise literature the SNR50 is also called the speech reception threshold in noise (SRT). It's not the same thing as the audiometric SRT, which is measured in quiet with two-syllable words.
Worked Example With Four-Talker Babble
Say a sentence test plays speech at a steady 63 dB while four-talker babble moves up and down. Your answers settle where the babble sits at 56.5 dB, so:
$$\text{SNR50} = 63 - 56.5 = +6.5\ \text{dB}$$Young adults with normal hearing average about +2 dB on sentence tests in speech babble, so your SNR loss is 6.5 − 2 = 4.5 dB, a mild loss. Now take a restaurant where your friend's voice is only 5 dB above the room. That's 1.5 dB worse than your 50 percent point. With a typical slope of about 10% per dB, you'd follow roughly 35% of the words, while a listener with normal hearing would get about 77%. If a hearing aid's directional microphone added 3 dB of effective SNR, the same table would move you to about +8 dB and roughly 65% of words.
Interpreting Speech-in-Noise Test Results
Your speech-in-noise test result makes the most sense against published categories. The table uses the SNR loss bands most clinics use for sentence tests. Digits-in-noise screening reports its own cut-offs, but the meaning of each band carries over.
| SNR loss | Category | What it usually means |
|---|---|---|
| 0 to 3 dB | Normal or near normal | You may need slightly better conditions than others in loud rooms |
| 3 to 7 dB | Mild | Busy restaurants are tiring; quieter seating helps |
| 7 to 15 dB | Moderate | Directional microphones on hearing aids are often recommended |
| Over 15 dB | Severe | Remote microphones or other assistive technology are usually needed |
As a reference point, people with mild-moderate sensorineural hearing loss average around 8 dB SNR-50 on a sentence test, about 6 dB worse than listeners with normal hearing.
Normative Data by Age Group
Speech-in-noise ability declines gradually with age, even when hearing thresholds look fine. Good tests publish normative data so your score can be read against your age group as well as against young adults. A score that counts as mild for a 25-year-old can be typical for someone in their late sixties, so the expected SRT range shifts upward by several dB across adulthood.
Use this context carefully. If you're one of many older adults with a borderline result, "normal for your age" doesn't mean "nothing to do". It means the difficulty is common, and the same hearing tools still help.
Percent Correct and the Psychometric Function
Some tests report percent correct at one fixed SNR instead of an SNR50. The two are linked by the psychometric function, an S-shaped curve of how many words you get right at each SNR. The SNR50 is where your curve crosses 50%, and the slope tells you how quickly you go from lost to comfortable.
The steep middle of that curve explains a common experience: a room that's only slightly louder can turn an easy conversation into guesswork. On the worked example's curve, moving from +8 dB to +5 dB takes you from about 65% of words to about 35%.
Checking Hearing Aid Programs With Three Rounds of SIN Tests
Daniel is 58, six weeks into his first pair of behind-the-ear hearing aids, and has a follow-up fitting in nine days. His question is narrow: is the "speech in noise" program his audiologist added worth switching to before Tuesday team meetings, or should he ask for a different setting? The answer, from his own numbers, is yes: the noise program gains him 1.4 dB, about 11 to 14 percentage points of words.
He sits at his kitchen table, sets his laptop volume once, and runs the digits-in-noise test three times in each condition, so a single lucky run can't decide anything:
| Condition | Runs (dB SNR) | Mean | Spread |
|---|---|---|---|
| Aids out | −2.9, −3.4, −3.1 | −3.13 | 0.5 dB |
| Default program | −4.3, −4.8, −4.4 | −4.50 | 0.5 dB |
| Noise program | −5.6, −6.2, −5.9 | −5.90 | 0.6 dB |
The arithmetic is plain subtraction. Aided benefit is −3.13 − (−4.50) = 1.37 dB on the default program and −3.13 − (−5.90) = 2.77 dB on the noise program. Because his own runs never spread more than 0.6 dB, I'd treat anything under about 1 dB as noise in the measurement; the 1.4 dB gap between programs clears that bar, the default program's benefit only just does. Using the published rule of thumb of 8 to 10 percentage points per dB near the 50% point, 1.4 dB is worth roughly 11 to 14 points of words in a meeting room.
The catch is the setup. With headphones or a laptop speaker, speech and noise reach his hearing aid microphones from the same direction, so directional processing can't separate them. His score here understates what a directional program can do in a real room, where the talker is in front and the clatter is behind. The number to trust is the difference between programs, not the absolute score.
His next step is specific: run the noise program at the next two meetings, write down how often he asks someone to repeat, and ask at the follow-up for a sound-field speech-in-noise test with babble from behind to measure the directional benefit properly.
Why Speech-in-Noise Testing Matters With a Normal Audiogram
The strongest case for speech-in-noise testing is that the audiogram doesn't predict it. Audiology research has shown this again and again. In a large study of veterans, about 70% scored 80% or better on word recognition in quiet, yet only 7% showed normal results on a words-in-noise task. One researcher called a speech-in-noise measure the stress test of hearing, because it pushes the auditory system the way a busy day does.
The test also has high face validity: you hear what you struggle with in real life, so the result is easy to explain and believe. That makes it a strong counseling tool. Seeing a number helps you set realistic expectations before you spend money on a hearing aid or other devices.
Hidden Hearing Loss and Cochlear Synaptopathy
Hidden hearing loss describes trouble in noise despite normal thresholds. One likely cause is cochlear synaptopathy: loss of the connections between inner-ear hair cells and the auditory nerve, often after noise exposure. The hair cells still respond to quiet tones, so the audiogram looks normal, but the nerve carries a less detailed signal. That detail is exactly what you need to separate a voice from babble. A speech-in-noise score is one of the few quick ways to show the problem exists.
Tinnitus, Attention and Listening Difficulties
Many people with tinnitus report listening difficulties in noise, even with near-normal hearing. The phantom sound takes up some of the same processing space as outside noise, so it acts like an extra layer of masking. Cognitive factors matter too: attention, working memory and processing speed all help you fill in words you only partly heard.
When results are poor but thresholds are normal, an audiologist may look at central auditory processing. A suspected auditory processing disorder needs a fuller test battery, but a speech-in-noise result is often what starts that conversation.
Why Word Recognition in Quiet Misleads
Classic word recognition tests use monosyllabic words, such as the NU-6 word list, read in a quiet booth. Many people score above 90% that way and still give up at a party. Scores in quiet simply don't predict scores in noise, which is why some researchers argue that word tests in noise should replace word tests in quiet.
Types of Speech-in-Noise Tests Clinics Use
Clinics have a range of speech-in-noise tests to choose from. They differ in what you listen to (digits, words or sentences), what the noise is, and how the SNR changes. The common ones in audiology clinics:
| Test | Speech material | Noise | Method | Result reported |
|---|---|---|---|---|
| QuickSIN | Sentences, female talker | Four-talker babble | Progressive, 25 to 0 dB | SNR loss in dB |
| Words in Noise (WIN) | NU-6 words | Multi-talker babble | Progressive, 24 to 0 dB | SNR50 in dB |
| Hearing in Noise Test (HINT) | Short sentences | Speech spectrum noise | Adaptive | SNR50 in dB |
| AzBio | Sentences, several talkers | Multi-talker babble | Fixed SNR | Percent correct |
| BKB-SIN | Simple sentences | Four-talker babble | Progressive | SNR loss in dB |
| LiSN-S | Sentences with spatial cues | Distracting stories | Adaptive | Spatial advantage in dB |
| Digits-in-noise | Digit triplets | Speech-shaped noise | Adaptive | SRT in dB SNR |
The QuickSIN is the most widely used because one list of six sentences takes about a minute, and most clinics run two QuickSIN lists and average them. The words-in-noise WIN test uses single words to limit how much memory and context help you. HINT was built to avoid ceiling and floor effects, and the LiSN-S is used for children with suspected spatial processing problems. The LiSN-S compares listening with the distracting talkers in front against talkers off to the side, so a small LiSN-S advantage points to a spatial listening issue rather than hearing loss. The Coordinate Response Measure is mostly a research tool for spatial hearing. In a clinic, the choice depends on the question: sentence tests show everyday function, while word and digit tests isolate the hearing side. Whichever test audiology staff pick, speech-in-noise testing works best when the same test is repeated over time.
Fixed, Adaptive and Progressive Methods
Every test changes the SNR in one of three ways:
- Fixed: one SNR for the whole test, such as +4 dB, scored as percent correct.
- Adaptive: the SNR moves up or down after each answer to home in on the 50% point, like this page's digits test.
- Progressive: the SNR steps down by a set amount whatever you answer, and the SNR50 is worked out from your score.
Tests also differ in which level moves. QuickSIN keeps speech steady and raises the background noise. The WIN test keeps the noise steady and lowers the speech.
Online Options Like EarCheck and the Word-in-Noise Test
Several online speech hearing tests exist besides this one. The Dutch EarCheck program is one of the best-studied digits-in-noise screeners. Other sites offer a word-in-noise check alongside a quiet speech reception test. Each uses its own norms, so compare your scores over time on the same site rather than across sites.
How Noise Type Changes a SIN Test
Two SIN test scores can differ just because the noise differs. The competing sound, the masker, can be steady or fluctuating, speech-like or not, and each type makes the task harder in a different way.
Babble vs Steady-State Noise
Steady-state noise such as white noise, pink noise or speech-shaped noise mainly causes energetic masking: it covers parts of the speech in the inner ear so they never reach you. Babble adds informational masking. The background voices sound like the one you want, so your brain has to sort which words belong to which talker.
With one to three background talkers you can catch words in the gaps, but you're also more easily confused. Beyond about four talkers the voices blur into a steadier wash. That's why four-talker babble is a common clinical choice: it sounds like a real room and still has plenty of informational masking.
Sentences, Words and Closed Sets
Sentences let you use context: if you catch "pass the salt and," you can guess "pepper." That makes sentence tests realistic but also sensitive to memory and language skills. Single words give fewer clues. A closed set, where you pick from a known list such as the digits 0 to 9, is easier than an open set, where any word could come next. That's why digits-in-noise scores look better on paper than sentence scores, and why each test needs its own norms.
SIN Testing for Hearing Aids and Cochlear Implant Users
SIN testing is more common in some groups than others, and hearing aid fitting is where speech-in-noise testing is most useful. For hearing aid users, a score taken before fitting shows how much help in noise to expect, and one taken after fitting shows whether the settings delivered it. For cochlear implant candidates, sentence tests in noise such as AzBio and BKB-SIN are part of the standard evaluation, and a poor score with well-fitted hearing aids is one sign that an implant may help more.
If you already wear hearing aids, take the online test twice, once with them in and once without. The difference shows how much each hearing aid helps you in noise, not just in quiet.
Aided vs Unaided Scores
Comparing aided and unaided scores tells you more than either one alone. Amplification restores audibility, which fixes the "louder please" part of hearing loss, but it can't fix distortion. If your SNR loss stays high once speech is audible, the bigger gains come from directional microphones, noise reduction settings and especially remote microphones that sit close to the talker, along with lip reading and good visual cues. Each dB of SNR you gain is worth about 8 to 10 percentage points of words, so the intervention that adds the most SNR matters most. Very easy tests can hit ceiling effects and hide that benefit, so clinics pick harder lists to measure aided results.
Online Hearing Screening vs a Clinical Speech-in-Noise Hearing Test
An online hearing screening and a clinical speech-in-noise hearing test answer different questions. This page tells you whether you might have a problem in noise and gives you a number to track. It isn't a clinical diagnostic test: your headphones, device and room are uncalibrated, and the result can't separate hearing loss from processing or attention problems.
If your result lands in the mild band or worse, or you score much worse than people your age, book a hearing evaluation with an audiologist or another hearing healthcare professional. Bring your score and ask for a speech-in-noise test as part of the full test battery. It adds only a few minutes to standard audiometry, and speech-in-noise testing supports a more accurate diagnosis and better communication advice.
Calibration, Sound Field and the Sound Booth
In a clinic, calibration sets the exact speech and noise levels in dB. Tests run either through headphones or in the sound field, with speech from a loudspeaker in front and babble from behind, in a sound booth or soundproof room. Sound-field testing is closer to real life and lets you wear your devices. Online tests can't control any of that, which is why your own listening trend over time on the same headphones counts for more than one absolute number.
Who Should Take Speech Hearing Tests
A quick speech hearing test is worth five minutes if:
- you follow speech well in quiet but get lost in groups, restaurants or meetings
- your hearing tests came back normal but listening in background noise still feels like hard work
- you have tinnitus and want to know how it affects understanding
- you're considering hearing aids, or want to check whether your current ones help in noise
- you want a baseline for your hearing health so you can retest each year
SIN tests are cheap and fast, and they measure the part of hearing you notice most: listening when everyone talks at once. Retest under the same conditions, watch the trend, and take a rising score to a professional sooner rather than later.
Frequently Asked Questions
- What does a speech-in-noise test measure?
- It measures how well you understand speech when other sound is playing at the same time. Instead of the softest beep you can detect in silence, it finds the signal-to-noise ratio at which you catch about half of what is said.
- Why can I hear in quiet but not in noisy restaurants?
- Following speech in noise relies on fine detail in the signal from your inner ear and on how your brain separates voices. Hidden hearing loss, auditory processing problems, tinnitus, attention and working memory can all make noisy rooms hard even when a standard hearing test is normal.
- How is this different from a regular hearing test?
- A regular hearing test plots the quietest tones you can hear in a silent booth. A speech-in-noise test adds competing noise and asks you to understand words, which is closer to the listening you do every day. The two results often disagree.
- What is a good speech-in-noise score?
- Lower is better. On this digits-in-noise test, a speech reception threshold of +2 dB SNR or lower falls in the normal range, +2 to +7 dB suggests mild difficulty, +7 to +15 dB moderate difficulty and above +15 dB significant difficulty. Compare it with the typical range for your age as well.
- Does tinnitus affect hearing in noise?
- It can. Many people with tinnitus find it harder to follow speech in background noise, because the internal sound competes with the voices you are trying to hear much like extra noise in the room.
- Is an online speech-in-noise test accurate?
- It is a screening, not a diagnosis. Your headphones, device and room are not calibrated, so use the score to spot a problem and track changes over time with the same setup, and see an audiologist for a clinical test such as QuickSIN if your result is mild or worse.
- Where can I get a clinical speech-in-noise test?
- Ask an audiologist during a hearing evaluation. A speech-in-noise test adds only a few minutes to a standard appointment, and you can request one if noisy places are your main complaint.