Attention
Digits-in-Noise Task
Also called: digits in noise, digit triplets in noise, speech-in-noise digit triplets, digit triplet recognition in noise, spoken digits in noise
Measure how well people make out spoken digits in noise: three digits from a synthetic voice, typed back, at seven fixed ratios of speech to noise.
9-12 minUntimed
Free to start: 10 free participants per account, then $0.99 per participant for academic research, or any yearly plan. Try it here first, with no account.
- Papers, 2021 to 2025
- 127
- Evidence grade
- B
- Registered clinical studies
- 15
What it measures
Measures how well people make out spoken digits in steady noise. Through headphones, a synthetic voice says three digits in noise shaped like speech, at seven fixed ratios of speech to noise, and participants type or tap the three digits. The main score is the share of triplets with all three digits right.
Constructs
- Speech perception in noise
- Auditory attention
- Recognition of spoken digits
Research uses
- Speech-in-noise perception research
- Aging and auditory cognition research, at the group level
- Within-study group comparison
How the paradigm works
The Digits-in-Noise Task measures how well people make out spoken digits in steady noise, the measure of the digits-in-noise test (Smits, Kapteyn and Houtgast, 2004; Smits, Theo Goverts and Festen, 2013): three digits play in noise shaped like speech, and a triplet counts only when all three digits are right. It needs headphones and normal hearing. The test starts with the sound setup: the digits and the noise load, the participant turns the device's volume until a steady tone is comfortable, and a six-trial headphone check (Woods, Siegel, Traer and McDermott, 2017) confirms that headphones are in use. By default a failed check ends the test, and the check's result is saved either way.
More about the paradigm
On each trial the participant presses Play and hears three different digits in noise, then types or taps them in order on a keypad and presses Done, with no time limit. The noise starts 500 ms before the first digit and ends 500 ms after the last, with 200 ms of noise alone between digits, as in De Sousa and colleagues (2020) and Polspoel and colleagues (2025); each trial's noise starts at a point drawn from the session's seed. Seven ratios of speech to noise are fixed before the session, from -18 to 0 dB in steps of 3 dB, each played 8 times (6 to 12 as a setting) in runs that hold every step once, in an order drawn from the seed: 56 triplets. Nothing adapts to the answers. Published versions move the ratio after each answer to find the point where half the triplets are right; fixed steps, as Polspoel and colleagues (2025) used to compare synthetic and natural digits, give the share right at each step. The noise plays 14 dB below the volume-setting tone at every step and the digits at that level plus the step, so at 0 dB they are as loud as the noise. Three practice triplets at 0 dB show the digits that played; the test gives no feedback.
The voices are synthetic: an open text-to-speech voice for each language, and noise with the long-term spectrum of that language's own ten digits. Every digit is set to the same speech level, but not adjusted for how easily it is heard in noise, as the digits of published versions are (Smits, Theo Goverts and Festen, 2013). In Japanese the digit 4 is left out until a listener has checked its clip, so Japanese triplets use nine digits. In Polspoel and colleagues (2025), 24 listeners with normal hearing made out synthetic and natural Dutch digits in noise with no significant difference.
The main score is the share of triplets right over all the steps, with the share at each step, the share of digits right in their places, and the ratio of speech to noise at which a curve fitted to the steps gives half the triplets right, with its slope. The scores describe how these synthetic digits were heard in this noise. The test gives no hearing level and no pass or fail, it is not a hearing test or a hearing screen, and no published result applies to its clips, so compare results within one language and within studies run the same way. Scores are research measures and do not diagnose.
Scores
- Triplets Right%
- Share of the triplets with all three digits right in their places, over all the steps played.
- Higher means more triplets were made out in the noise
- Triplets Right by Step%
- Share of triplets right at each ratio of speech to noise, keyed by the ratio in dB.
- Rises with the ratio
- Digits Right%
- Share of the digits right in their places, over all the steps.
- Higher means more digits were made out; guessing gives about 0.1
- Ratio for Half the TripletsdB
- Ratio of speech to noise at which a cumulative Gaussian fitted to the triplets right at each step gives half the triplets right.
- Lower means fewer digits were needed above the noise; empty outside -18 to 0 dB
- Slope at Halfscore
- Slope of the fitted curve at that point, in percentage points per dB.
- Informational
- Median Answer Timems
- Median time from when the keypad opened, after the digits, to Done.
- Informational; the answer has no time limit
- Anticipationscount
- Answers whose first digit came within 100 ms of the keypad opening, which leave the scores.
- Half or more of the answers makes the session invalid
- Completed Trialstrials
- Triplets answered in the scored run.
- Fewer than planned is flagged for review
- Planned Trialstrials
- Triplets the run set out to play.
- Informational
- Headphone Check Passed0/1
- 1 when the headphone check passed (at least 5 of 6 right), 0 when it did not.
- 0 stops the test unless the study set it to continue
- Headphone Check Correctcount
- Headphone check trials answered right, of 6.
- 5 or 6 passes
- Reported Output Delayms
- Output delay the device reported to the browser; empty when the browser does not report it.
- Informational
Use and evidence
127 papers from 2021 to 2025 name this task in their title or abstract, up 127% compared with 2016 to 2020. That ranks it 188 of 495 test families in our census. See the counts for every test.
Show the numbers
| Year | Papers |
|---|---|
| 2025 | 40 |
| 2024 | 28 |
| 2023 | 21 |
| 2022 | 17 |
| 2021 | 21 |
| 2020 | 18 |
| 2019 | 13 |
| 2018 | 11 |
| 2017 | 7 |
| 2016 | 7 |
| 2015 | 0 |
| 2014 | 4 |
| 2013 | 2 |
| 2012 | 1 |
| 2011 | 0 |
| 2010 | 0 |
| 2009 | 0 |
| 2008 | 0 |
| 2007 | 0 |
| 2006 | 0 |
| 2005 | 0 |
| 2004 | 0 |
| 2003 | 0 |
| 2002 | 0 |
| 2001 | 0 |
| 2000 | 0 |
Use grade C
100 to 299 papers from 2021 to 2025.
Evidence grade B
At least 8 test-retest papers, and either 5 registered clinical studies or 150 reliability or validity papers.
- Papers on test-retest reliability
- 25
- Papers on reliability or validity
- 66
- Registered clinical studies using it as an outcome
- 15
How the grades work
Use: A is 1,000 or more papers from 2021 to 2025, B 300 to 999, C 100 to 299, D 30 to 99 and E fewer than 30. Evidence: A means at least 25 papers report its test-retest reliability and at least 25 registered clinical studies use it as an outcome; B, at least 8 test-retest papers, and either 5 registered clinical studies or 150 reliability or validity papers; C, at least 2 test-retest papers, 30 reliability or validity papers or 3 registered clinical studies; D, little published measurement evidence found. The counts measure use. They include every paper that names the task in its title or abstract. Grades describe the published research on this task, not ConductCognition's version of it.
Source: ConductCognition test census 0.22.0, counted 31 October 2026 from OpenAlex and ClinicalTrials.gov.
Task design
The trial sequence as the test runs in ConductCognition.
- Sound setup
- The digits and the noise load, the participant turns the device's volume until a steady 1 kHz tone is comfortable, then takes a 6-trial headphone check (Woods and colleagues, 2017). By default a failed check ends the test; its result is saved either way.
- Digits
- The digits 0 to 9 said by a synthetic voice in each language, from the platform's own speech clips, all set to one speech level. Japanese leaves out 4 until a listener has checked its clip.
- Noise
- Noise with the long-term spectrum of the language's ten digits, from 500 ms before the first digit to 500 ms after the last, with 200 ms of noise alone between digits. Each trial's noise starts at a point drawn from the session's seed and fades in and out over 50 ms.
- Trial
- Press Play to hear three different digits in noise, then type or tap them in order on a keypad and press Done, with no time limit.
- Steps
- Seven ratios of speech to noise, -18, -15, -12, -9, -6, -3 and 0 dB, set before the session: the noise 14 dB below the volume-setting tone and the digits that level plus the step. No answer changes the next step: there is no staircase.
- Trials
- Each step 8 times (standard, 6 to 12), in runs that hold every step once, in an order drawn from the session's seed: 56 triplets, with no feedback. Every digit is used as evenly as the count allows, and no triplet comes twice.
- Practice
- 3 triplets at 0 dB, showing the digits that played for 1,500 ms.
- Score
- The share of triplets with all three digits right, over all the steps and at each one, with the share of digits right in their places and the ratio at which a cumulative Gaussian fitted by maximum likelihood gives half the triplets right. Answers whose first digit came under 100 ms leave the scores.
The voices are synthetic, made for this platform, and their digits are not adjusted for how easily each is heard in noise as published versions' digits are, so no published result applies; compare results within one language and within studies run the same way. The test needs normal hearing and headphones. It is not a hearing test: it gives no hearing level and no pass or fail, and it does not screen hearing. The scores support research and do not diagnose.
Quality checks
Each session is marked valid, needs review or invalid, with the reason, before you see its scores.
Scored rows: digits_in_noise_test from the scored run, three digits typed or tapped after the sound has played, with no time limit, graded from the digits played and the digits entered: a triplet is right only with all three in place (chance 1 in 1,000 for a guess on the keypad). Below chance invalidates and not above chance is flagged; the same three digits on 90% of answers invalidates when accuracy is not above chance; first digits under 100 ms from the keypad opening invalidate when they are at least half. A headphone check that stopped the test invalidates; a failed check with the test run anyway, fewer than half the triplets right at the easiest step, no fitted 50 percent point, or a run short of its planned trials is flagged for review.
Why these rules
Accuracy significantly below chance (one-sided exact test, p < .05) cannot come from guessing; it means the response mapping was reversed or the task was not followed (Slick, Sherman and Iverson 1999 treat below-chance forced-choice performance as definite response bias). Chance-level accuracy can be a real result in impaired participants, so it is only flagged. One option on at least 90% of answered trials, when the correct options are balanced and accuracy is not above chance, means the answers did not depend on the items (the task counterpart of long-string responding in surveys; Curran 2016). The keypad opens once the digits have played and has no time limit, so a first digit under 100 ms was pressed before it opened, and only a session made mostly of such answers is invalid. Woods, Siegel, Traer and McDermott (2017) let only listeners who passed the headphone check go on, because a listener on loudspeakers hears the sounds differently; a session that ran after a failed check keeps its scores but not its percentile. At 0 dB, the easiest step, the digits are as loud as the noise, so most triplets should be right when the sound plays as intended.
No percentile. Scores and quality flags only.
Running this test
Settings you can change
| Setting | Standard | Allowed |
|---|---|---|
| When the headphone check fails | Stop the test | Stop the test, Continue the test |
| Triplets per step | 8 | 6 to 12 |
Yearly plans can change settings within these limits. Free and pay-as-you-go studies use the standard settings.
Need a setting this test does not have?
Devices and languages
- Untimed test: runs on desktops, laptops, tablets and phones. Needs headphones. Participant devices
- Languages: English, Spanish, French, German, Italian, Portuguese (Brazil), Chinese (Simplified) and Japanese.
- Available on: Every plan, including the free plan.
- No percentile. Scores and quality flags only. How norms work
- Recruiting online? See the setup guides for Prolific, Sona Systems, CloudResearch Connect and Qualtrics.
Cite and describe
Participants completed the Digits-in-Noise Task in a web browser (ConductCognition task version 1.0.0, scoring version 1.0.0; Smits, Kapteyn and Houtgast, 2004; Smits, Theo Goverts and Festen, 2013), with the standard settings (when the headphone check fails: Stop the test; triplets per step: 8). The main score was triplets right (share of the triplets with all three digits right in their places, over all the steps played).
ConductCognition versions: task version 1.0.0, scoring version 1.0.0. Every export records the versions each session used, for a methods section.
- 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. https://doi.org/10.1080/14992020400050004Digit triplets in noise, a triplet counted only with all three digits right.
- Smits, C., Theo Goverts, S., & Festen, J. M. (2013). The digits-in-noise test: Assessing auditory speech recognition abilities in noise. The Journal of the Acoustical Society of America, 133(3), 1693-1706. https://doi.org/10.1121/1.4789933Digits in noise with the long-term spectrum of speech, and digits made equally easy to hear by level corrections.
- Koopmans, W. J. A., Goverts, S. T., & Smits, C. (2018). Speech recognition abilities in normal-hearing children 4 to 12 years of age in stationary and interrupted noise. Ear and Hearing, 39(6), 1091-1103. https://doi.org/10.1097/AUD.0000000000000569Unique triplets built from the digits 0 to 9, with noise before and after each triplet.
- 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. https://doi.org/10.1097/AUD.0000000000000775500 ms of noise before and after a triplet, 200 ms between digits, and fresh noise on each trial from a random point in a long noise.
- Polspoel, S., Moore, D. R., Swanepoel, D. W., Kramer, S. E., & Smits, C. (2025). Automatic development of speech-in-noise hearing tests using machine learning. Scientific Reports, 15, 12878. https://doi.org/10.1038/s41598-025-96312-zSynthetic digits in noise at fixed ratios, triplets of three different digits, and no significant difference between synthetic and natural digits.
- Van den Borre, E., Denys, S., van Wieringen, A., & Wouters, J. (2021). The digit triplet test: A scoping review. International Journal of Audiology, 60(12), 946-963. https://doi.org/10.1080/14992027.2021.1902579The versions of the paradigm across languages, noises and procedures.
- Woods, K. J. P., Siegel, M. H., Traer, J., & McDermott, J. H. (2017). Headphone screening to facilitate web-based auditory experiments. Attention, Perception, & Psychophysics, 79(7), 2064-2072. https://doi.org/10.3758/s13414-017-1361-2The headphone check before the test.
Source papers
- Polspoel S, Moore DR, Swanepoel DW, Kramer SE, Smits C (2025). Automatic development of speech-in-noise hearing tests using machine learning. Scientific Reports, 15:12878.N 24, Young adults
- Smits C, Theo Goverts S, Festen JM (2013). The digits-in-noise test: assessing auditory speech recognition abilities in noise. The Journal of the Acoustical Society of America, 133(3):1693-1706.N 1386, 60+
Questions
How long does the Digits-in-Noise Task take?
About 9 to 12 minutes, as listed in the test library.
What device does a participant need?
Untimed test: runs on desktops, laptops, tablets and phones. Needs headphones. The browser must be Chrome, Edge, Firefox or Safari.
What scores does it produce?
Triplets Right, Triplets Right by Step, Digits Right, Ratio for Half the Triplets, Slope at Half, Median Answer Time, Anticipations, Completed Trials, Planned Trials, Headphone Check Passed, Headphone Check Correct and Reported Output Delay. Each session is also marked valid, needs review or invalid, with the reason.
How widely is the Digits-in-Noise Task used in research?
127 papers from 2021 to 2025 name it in their title or abstract, up 127% compared with 2016 to 2020. That ranks it 188 of 495 test families in the ConductCognition test census (use grade C). Source: ConductCognition test census 0.22.0, counted 31 October 2026 from OpenAlex and ClinicalTrials.gov.
How much reliability evidence does the Digits-in-Noise Task have?
In the published research, 25 papers report its test-retest reliability and 15 registered clinical studies use it as an outcome (evidence grade B in the ConductCognition test census).
Does it report a percentile?
No. Scores and quality flags only.
Which settings can a researcher change?
When the headphone check fails (Stop the test, Continue the test) and triplets per step (6 to 12). Settings can be changed on yearly plans; free and pay-as-you-go studies use the standard settings.
Which languages can participants use?
Participants can take this test in English, Spanish, French, German, Italian, Portuguese (Brazil), Chinese (Simplified) and Japanese, with its words in the same language. English is on every plan, and the other languages come with yearly plans. Have a native speaker read the machine-translated text before launch.
Is this a clinical test?
No. ConductCognition is for research use only. It runs, scores and exports the test; clinical interpretation is outside the platform.
Every test: the full test list.
Run this test in your next study
Start free with 10 free participants on any test, then use participant credits or a yearly plan.