Attention
Timbre Task
Also called: timbre dissimilarity task, timbre perception task, timbre discrimination task, timbre space ratings, timbre similarity ratings, tone color discrimination
Measure how people hear timbre: synthetic tones of one pitch that differ in brightness, attack and spectral flux, rated in pairs for how different they sound, then an odd one out of three.
7-9 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
- 79
- Evidence grade
- C
- Registered clinical studies
- 3
What it measures
Measures how people hear timbre, the quality that tells two instruments apart on the same note. Synthetic tones of one pitch, loudness and length differ in brightness (spectral centroid), attack and spectral flux. Participants rate how different pairs of tones sound, then pick the odd one of three. Scores give how much each feature shapes the ratings, and the odd-one-out accuracy; across a study, the ratings support multidimensional scaling. Needs headphones and normal hearing; results are research measures, not a hearing test.
Constructs
- Timbre perception
- Auditory discrimination
- Spectral and temporal sound features
Research uses
- Music perception research
- Auditory perception research
- Within-study group comparison
How the paradigm works
Timbre is the quality that tells two instruments apart when they play the same note at the same loudness. Dissimilarity ratings of tones equal in pitch, loudness and length, analyzed by multidimensional scaling, give a timbre space (Grey, 1977), whose common dimensions follow the attack, the spectral centroid and the change of the spectrum over time (McAdams and colleagues, 1995); synthetic tones that vary these features one at a time show which ones listeners use (Caclin and colleagues, 2005). This task does both in one session: ratings of every pair of eight synthetic tones, then an odd-one-out test of each feature.
More about the paradigm
Set a comfortable volume and complete the 6-trial headphone check. A failed check stops the task by default; researchers can choose to continue with a quality flag.
Tones of 600 ms, 20 harmonics of 311.13 Hz, that differ on three features: the spectral centroid (brightness) from harmonic 3 to 6, a linear attack from 15 to 200 ms, and spectral flux, a glide of the centroid by up to an octave over the tone. Each tone ends with a 100 ms fall. Levels are matched by A-weighted rms, 12 dB below the volume-setting tone.
Ratings: the eight tones at the ends of the three features, every pair once and each tone with itself: 36 pairs, 400 ms apart, rated from 1 (the same) to 7 (very different), after 2 practice pairs.
Odd one out: the tone at the middle of all three features twice, and once a tone 1/8, 1/4 or 1/2 of one feature's range away, up or down; choose First, Second or Third. 4 trials of each feature and step (2 to 8 as a setting): 36 trials, after 3 practice trials with feedback for 1,000 ms. Each tone's level moves at random by up to 2 dB, so loudness is no guide.
Scores: odd-one-out accuracy overall, by feature and by step; the mean rating of pairs of one tone and of two different tones; and each feature's weight in the ratings, from a least-squares fit of the ratings on which features differ, with the share of their variance it explains. Across a study, the ratings feed multidimensional scaling such as CLASCAL or INDSCAL. Responses under 100 ms from the answer buttons opening leave the scores.
Scores
- Odd-One-Out Accuracy%
- Share of odd-one-out trials answered with the tone that differed.
- Higher means more timbre differences heard; chance is one third
- Brightness Accuracy%
- Odd-one-out accuracy when the odd tone differed in spectral centroid.
- Higher means more brightness differences heard
- Attack Accuracy%
- Odd-one-out accuracy when the odd tone differed in attack.
- Higher means more attack differences heard
- Flux Accuracy%
- Odd-one-out accuracy when the odd tone differed in spectral flux.
- Higher means more flux differences heard
- Accuracy by Step%
- Odd-one-out accuracy at each step, keyed by its share of a feature's range: 0.125, 0.25 and 0.5.
- Rises with the step
- Different Pairs Ratingpoints
- Mean rating, 1 to 7, of pairs of two different tones.
- Informational; there is no correct rating
- Same Pairs Ratingpoints
- Mean rating, 1 to 7, of pairs that played one tone twice.
- Near 1 is expected; as high as the different pairs is flagged
- Brightness Weightpoints
- Rating points added when the two tones differ in spectral centroid, from a least-squares fit of the ratings on which features differ.
- Higher means brightness shaped the ratings more
- Attack Weightpoints
- Rating points added when the two tones differ in attack, from the same fit.
- Higher means attack shaped the ratings more
- Flux Weightpoints
- Rating points added when the two tones differ in spectral flux, from the same fit.
- Higher means flux shaped the ratings more
- Ratings Fit%
- Share of the ratings' variance that the three features explain (R squared).
- Higher means the ratings follow the tones' features more closely
- Median Answer Timems
- Median time from when the answers opened, after the tones, to the answer, over both parts.
- Informational; the answer has no time limit
- Anticipationscount
- Answers within 100 ms of the answers opening, which leave the scores.
- Half or more of the answers makes the session invalid
- Completed Trialstrials
- Ratings and odd-one-out trials answered in the scored run.
- Fewer than planned is flagged for review
- Planned Trialstrials
- Ratings and odd-one-out trials 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
79 papers from 2021 to 2025 name this task in their title or abstract, down 15% compared with 2016 to 2020. That ranks it 318 of 670 test families in our census. See the counts for every test.
Show the numbers
| Year | Papers |
|---|---|
| 2025 | 20 |
| 2024 | 18 |
| 2023 | 10 |
| 2022 | 16 |
| 2021 | 15 |
| 2020 | 10 |
| 2019 | 27 |
| 2018 | 21 |
| 2017 | 13 |
| 2016 | 22 |
| 2015 | 11 |
| 2014 | 14 |
| 2013 | 28 |
| 2012 | 22 |
| 2011 | 13 |
| 2010 | 13 |
| 2009 | 15 |
| 2008 | 11 |
| 2007 | 10 |
| 2006 | 11 |
| 2005 | 4 |
| 2004 | 4 |
| 2003 | 7 |
| 2002 | 6 |
| 2001 | 5 |
| 2000 | 9 |
Use grade D
30 to 99 papers from 2021 to 2025.
Evidence grade C
At least 2 test-retest papers, 30 reliability or validity papers or 3 registered clinical studies.
- Papers on test-retest reliability
- 2
- Papers on reliability or validity
- 10
- Registered clinical studies using it as an outcome
- 3
Measured on ConductCognition
Pooled across the studies that feed our platform norms, by device group. A figure needs 50 or more people.
No pooled figure yet.
Note on the counts: includes music-technology papers.
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 0.48.0, counted 12 November 2026 from OpenAlex and ClinicalTrials.gov.
Task design
The trial sequence as the test runs in ConductCognition.
- Sound setup
- 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.
- Tones
- 600 ms tones of 20 harmonics of 311.13 Hz, with the same pitch, length and A-weighted level, 12 dB below the volume-setting tone. They differ on three features: brightness (the spectral centroid, from harmonic 3 to 6), attack (a rise of 15 to 200 ms) and spectral flux (a glide of the centroid by up to an octave over the tone).
- Ratings
- The eight tones at the ends of the three features, every pair once and each tone with itself: 36 pairs, 400 ms apart, each rated from 1 (the same) to 7 (very different) with no time limit, after 2 practice pairs.
- Odd one out
- Three tones: the tone at the middle of all three features twice, and once a tone 1/8, 1/4 or 1/2 of one feature's range away. Choose First, Second or Third. Each tone's level moves at random by up to 2 dB, so loudness is no guide.
- Trials
- 4 odd-one-out trials of each feature and step (standard, 2 to 8): 36 trials, with no feedback, after 3 practice trials with feedback for 1,000 ms.
- Score
- Odd-one-out accuracy overall (chance 1 in 3), by feature and by step. From the ratings: the mean rating of pairs of one tone and of two tones, and each feature's weight, the rating points it adds, from a least-squares fit with the share of variance it explains. Answers under 100 ms leave the scores.
- Across a study
- Each rating is saved with its pair of tones and their features, ready for multidimensional scaling such as CLASCAL or INDSCAL, which recovers the timbre space from many listeners' ratings.
The tones play through the participant's own headphones at a volume they chose, and loudness is matched by an A-weighted approximation, so compare results within studies run the same way. Pairs of one tone rated as different as pairs of two tones flag a session. The test needs normal hearing and headphones, wired if possible: wireless headphones can add delay and change the sound. It is a research measure, not a hearing or musical-ability test: it gives no hearing level and no pass or fail. No percentile is given.
Quality checks
Each session is marked valid, needs review or invalid, with the reason, before you see its scores.
Scored rows: timbre_pairs_rating_test and timbre_pairs_odd_test from the selected run: ratings from 1 to 7 of how different two tones sound, then First, Second or Third for the odd tone of three, each after the tones have played, with no time limit. Odd-one-out answers are graded from the saved button and the odd tone's place, chance one third; ratings have no correct answer. Below chance invalidates and not above chance is flagged; one answer on 90% of trials invalidates when accuracy is not above chance; answers under 100 ms leave the scores and invalidate when at least half. Pairs of one tone rated as different as pairs of two tones, a failed headphone check with the test run anyway and a run short of its planned trials are flagged for review; a check that stopped the test leaves no results.
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). Ratings have no correct answer, but a listener who hears the tones rates a tone with itself as more alike than two different tones. The answers open once the tones have played, so an answer under 100 ms was pressed before they opened. 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.
No percentile. Scores and quality flags only.
Running this test
Settings you can change
| Setting | Standard | Allowed |
|---|---|---|
| Odd-one-out trials per step | 4 | 2 to 8 |
| When the headphone check fails | Stop the test | Stop the test, Continue the test |
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 Timbre Task in a web browser (ConductCognition task version 1.0.0, scoring version 1.0.0; Grey, 1977; McAdams, Winsberg, Donnadieu, De Soete and Krimphoff, 1995; Caclin, McAdams, Smith and Winsberg, 2005), with the standard settings (odd-one-out trials per step: 4; when the headphone check fails: Stop the test). The main score was odd-one-out accuracy (share of odd-one-out trials answered with the tone that differed).
ConductCognition versions: task version 1.0.0, scoring version 1.0.0. Every export records the versions each session used, for a methods section.
- Grey, J. M. (1977). Multidimensional perceptual scaling of musical timbres. The Journal of the Acoustical Society of America, 61(5), 1270-1277. https://doi.org/10.1121/1.381428A timbre space from dissimilarity ratings of tones equal in pitch, loudness and length.
- McAdams, S., Winsberg, S., Donnadieu, S., De Soete, G., & Krimphoff, J. (1995). Perceptual scaling of synthesized musical timbres: Common dimensions, specificities, and latent subject classes. Psychological Research, 58(3), 177-192. https://doi.org/10.1007/BF00419633Common dimensions of timbre: log rise time, spectral centroid and spectral variation.
- Caclin, A., McAdams, S., Smith, B. K., & Winsberg, S. (2005). Acoustic correlates of timbre space dimensions: A confirmatory study using synthetic tones. The Journal of the Acoustical Society of America, 118(1), 471-482. https://doi.org/10.1121/1.1929229Synthetic tones varying attack time, spectral centroid and spectral flux; flux the least salient.
- Lee, H., & Müllensiefen, D. (2020). The Timbre Perception Test (TPT): A new interactive musical assessment tool to measure timbre perception ability. Attention, Perception, & Psychophysics, 82(7), 3658-3675. https://doi.org/10.3758/s13414-020-02058-3Individual differences in timbre perception, measured with synthetic tones.
- 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
- Grey (1977): a timbre space from dissimilarity ratings. McAdams and colleagues (1995): its common dimensions and classes of listeners. Caclin and colleagues (2005): attack time, spectral centroid and spectral flux varied one at a time in synthetic tones.Paradigm source, Adults
Questions
How long does the Timbre Task take?
About 7 to 9 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?
Odd-One-Out Accuracy, Brightness Accuracy, Attack Accuracy, Flux Accuracy, Accuracy by Step, Different Pairs Rating, Same Pairs Rating, Brightness Weight, Attack Weight, Flux Weight, Ratings Fit, 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 Timbre Task used in research?
79 papers from 2021 to 2025 name it in their title or abstract, down 15% compared with 2016 to 2020. That ranks it 318 of 670 test families in the ConductCognition test census (use grade D). Source: ConductCognition test census 0.48.0, counted 12 November 2026 from OpenAlex and ClinicalTrials.gov.
How much reliability evidence does the Timbre Task have?
In the published research, 2 papers report its test-retest reliability and 3 registered clinical studies use it as an outcome (evidence grade C in the ConductCognition test census).
Does it report a percentile?
No. Scores and quality flags only.
Which settings can a researcher change?
Odd-one-out trials per step (2 to 8) and when the headphone check fails (Stop the test, Continue the test). 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. 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.