Sound recognition questions play a clip and ask what made it: which instrument, which animal or bird, which letter sound, which murmur, which word. Nothing on the page describes the sound, so the answer can only come from the ear.
EdProcto, the quiz maker that publishes this guide, carries the shape as Recognise the sound: a clip, a prompt and options, marked by rule. The guide is for anyone who teaches by ear, in music, early reading, languages, medicine or natural history.
We read each source on 10 October 2026: Ogawa et al. (2023) and Lively et al. (1993) in full, the other studies as abstracts. Questions on a longer recording have their own guide.
What does a sound recognition question measure that a written one cannot?
It measures whether a learner can pick the sound out by ear, which is different from knowing about it. A student can describe the murmur of a narrowed valve from the textbook without ever having recognised one.
Medicine has the starkest figures. Mangione and Nieman (1997) played 12 cardiac events, recorded directly from patients, to 453 physicians in training and 88 medical students, who named each from a multiple-choice list. The residents recognised 20% on average, improved little with each year of training and did not score significantly higher than the students.
So recognising is a skill of its own, and only a question that plays the sound tests it. What it leaves out is the meaning behind the name: set a written question beside it for that.
What did research find about learning to recognise sounds by testing?
It found that people learn sounds fast from repeated hearing and from naming them with feedback, that mixed practice and varied recordings help the learning carry over, and that learners misjudge which practice works.
| Field | Study | What was found |
|---|---|---|
| Birdsong | Ogawa et al. (2023): 66 students, 26 species, 200 quiz questions over four days | Scores rose from 3.65 to 14.62 out of 26, and stood at 12.45 two weeks later |
| Heart sounds | Barrett et al. (2004): 51 medical students, 500 hearings of each of four basic murmurs | Recognition rose from 13.5% to 85% in one group and from 20.9% to 86.1% in another; ten untrained students went from 24% to 32% |
| Music (ear training) | Wong et al. (2021): novices, twelve kinds of melodic interval | Mixed practice beat one kind at a time, when no reference songs were given |
| Speech sounds | Lively et al. (1993): Japanese listeners, English r and l | Training on five voices carried over to a new voice; training on one did not |
| Letter sounds | Ehri et al. (2001): 52 studies of teaching children the sounds in words | Effects of d = 0.86 on phoneme awareness and 0.53 on reading, larger when letters were used |
No study we read set quizzing on sounds against plain listening: these show that both work, not which is better.
Quizzing with feedback: birdsong and speech sounds
In Ogawa et al. (2023) each practice question played a recording and offered five species; the answer screen named the right one, with its photograph, a picture of its sound and every option's sound to hear. The authors note that no group went untrained and that tests and practice drew on one source of recordings.
One result surprised them: a version that asked mostly about the species a learner kept missing gave lower scores than questions drawn at random. They suspect it grew hard faster than beginners could follow, so early on keep known sounds in the set.
Lively et al. (1993) trained listeners for fifteen days. Each trial played one word of a pair differing only in r or l; a light showed the right answer, and a missed word was played again. The six trained on five speakers identified new words from a new speaker as accurately as from a familiar one, 88% each. Five others, trained on one speaker, improved without carrying the gain to a new voice.
Mixing and repetition: intervals, heart sounds and letter sounds
Wong et al. (2021) had novices learn six kinds of interval mixed together and six one kind at a time. With familiar reference songs and singing as aids the two schedules came out alike; without them, mixing gave clearly better identification of new examples, though most participants judged blocking the better. Our interleaving guide has the wider evidence.
Barrett et al. (2004) tested plain repetition, and conclude that listening to the heart is partly a technical skill. Ehri et al. (2001) found the teaching of sounds more effective with letters than without, so play the sound and offer letters; our guide for children who cannot read yet has more.
How do you choose and cut the clip?
Choose a recording in which the sound stands alone, and cut it so that nothing except the sound can give the answer.
- One sound a clip. A field recording with a second bird behind the first has two right answers: cut to where one sings alone. For a heart, one sound is one finding over many beats: at 75 beats a minute, ten seconds hold about twelve.
- Keep the opening. Siedenburg (2019) trained listeners to name ten instruments, then tested them on slices of 64 milliseconds: accuracy fell by 25% when the slice came from the middle of the note and not its start. Begin each clip just before the sound.
- Keep it short. Our working rule, not a finding: the sound heard whole, twice or three times, in under fifteen seconds.
- Level the loudness. If the trumpet is the loud clip, loudness becomes the clue. Audacity's manual (Audacity Team, 2026) describes an effect, Loudness Normalization, that brings a track to a set perceived loudness. Put every clip of a set through one setting.
- Remove other clues. No spoken introduction, and files named by number: the clip's address is what the student's browser fetches.
- Record each thing more than once. One oboe recording teaches that recording; several players, birds or voices teach the sound, as five speakers did in Lively et al. (1993).
- Build it and sit it. In EdProcto, choose Recognise the sound, paste the clip's address, and sit the quiz yourself on a phone, with ordinary earphones.
Which wrong options should you offer?
Offer the sounds a learner really mixes up, because an option nobody could mistake for the clip tests nothing.
| The clip | Wrong options | Why these |
|---|---|---|
| A solo phrase on the oboe | Clarinet and cor anglais, before trumpet or drum | Confusions are significantly more frequent within an instrument family (Giordano and McAdams, 2010; abstract read) |
| The word "lock" | "rock" | A pair one sound apart, as in Lively et al. (1993); the words are our example |
| Twelve beats with one murmur | Other murmurs, one at the same point in the beat | Mangione and Nieman (1997) adjusted scores when acoustically similar findings were picked |
| One bird, one phrase | Birds heard in the same place and season | Ogawa et al. (2023) chose the 26 species commonest in their survey recordings |
| "mmm", said once | n and b, before s | Our suggestion: sounds made in a similar way |
EdProcto takes from two to eight options; Ogawa et al. offered five and Lively et al. two. Three real confusions beat six with padding, and how to write distractors has the general rules. After a sitting, the full analysis on EdProcto's Educator plan counts the takers of each option: replace one nobody took, and treat one that drew half the class as your next lesson.
How many times should the clip play?
As often as the learner likes in practice, and a fixed number in a test, set by how the sound is met in life.
| The sound | Hearings | In EdProcto's Recognise the sound |
|---|---|---|
| Any sound, in practice | Without limit: Barrett et al. (2004) used 500 hearings a murmur, and Lively et al. (1993) replayed each missed word | Leave "Plays once, no replay" off: the browser's own player pauses, goes back and replays |
| In a test, a sound that goes on: a heartbeat | Replays, as under the stethoscope | The same |
| In a test, a sound that passes: a bird overhead, a letter said in class | One, or two | Switch it on; for two hearings, record the sound twice in one clip, a second of silence between |
With the switch on, the student has a single button, "Play the clip", a line that fills as the clip runs, and the words "It plays once, from start to end: there is no pause and no going back." For longer recordings, see once or twice.
Which clips do you have the right to use?
Your own recordings, and anyone else's only on the terms its owner sets. This is not legal advice: ask whoever handles copyright where you work.
| Source | What its own page says | So |
|---|---|---|
| Your own recording | Nothing to read | Ask anyone whose voice is on it; a patient's heart sounds go through your institution's consent process |
| University of Iowa Musical Instrument Samples (Fritts, n.d.) | Notes and scales that "may be downloaded and used for any projects, without restrictions" | Single notes for instrument questions |
| Freesound (n.d.) | Each sound has its own licence: almost no conditions (CC0), credit required, or credit and non-commercial use only | Read the licence of every sound you take |
| Any Creative Commons work | Creative Commons (n.d.): BY asks for credit, NC allows non-commercial use only, ND permits no adaptations, SA puts adaptations under the same terms | Under ND, ask before cutting an excerpt; under NC, before use in a course you sell |
An excerpt of a commercial recording is not allowed merely for being short. Where a licence asks for credit, a credit that names the bird answers the question: put it in the explanation, shown only after answering, or on your course page.
What do you do for a student who cannot hear the clip?
Decide whether hearing is the thing being tested; if it is, set another task in the same subject for the same marks, agreed before the quiz.
Where a clip only carries words, such as a spoken instruction, give the words in print to everyone. Where hearing is the skill, a description ("an oboe plays") is the answer, so text cannot stand in: our mixed media guide sets out what the web's accessibility guidance asks, and its exception for tests.
Trinity College London (n.d.) gives candidates with hearing loss an aural awareness test in place of its standard aural test: the examiner taps the rhythm of a short melody twice on a table top, and the candidate, watching and feeling the vibration through a hand on the table, claps it back. Trinity asks for evidence of need and keeps its assessment standards the same for all candidates. The subject is kept and the sense is changed.
For a class quiz, with your institution's policy and the law first, either set another route to the same aim (a rhythm tapped, the sound's trace or a score to read), recording that it tested reading the sound and not hearing it, or leave the sound questions out for that student and mark the rest. In EdProcto, do either in a copy made with "Duplicate this quiz", which has a code of its own. The "Accessibility" switches help with reading and seeing, not with hearing, so the other route is yours to set.
A student whose earphones fail is a different case. Open the quiz with a sound check: a clip of your voice, the prompt "Can you hear this clearly?", and the switch "Not marked: a reflection, not a question". An unmarked opening question keeps first place when the rest are shuffled.
How does EdProcto run a sound recognition question?
EdProcto is the best tool for sound recognition questions from your own recordings, because the clip, the options and the number of hearings are set on the question itself and the marking is by rule. As a study quiz it checks each answer as it is given, the name-it-and-be-told practice of the birdsong and speech studies; as an exam it marks at hand-in.
- Building. Recognise the sound is in the builder now, among the media prompts. With "It is" at "Audio", the box "Audio clip address" takes the web address of the file, which is handed to the student's browser as typed: it must open the sound file itself, without a sign-in. "Prompt" takes the question, and under "Options · mark the right one" you type the names and mark the key.
- Afterwards. "Explanation shown to the student afterwards" is the place for what to listen for next time, and for the credit.
- Neighbouring shapes. Match the sound offers pictures as the options, for children who cannot read the names, and Listen once has the one-play rule built in. Pick the clip, where the options are themselves clips, and Compare two clips are marked by rule too.
A second phone can record any clip that plays, so where the result counts, add the integrity settings the stakes call for.
The Free plan has ten quizzes a month for thirty students each, on the core templates, with marking by rule. Tutor opens all 111 templates for 100 students a quiz, and Educator adds the full analysis for 500. See pricing.
Quick questions
How long should a sound clip be?
Long enough to hear the sound whole two or three times, usually under fifteen seconds: our working rule. Keep the start of the sound (Siedenburg, 2019).
How many times should students hear the clip?
Without limit in practice. In a test, as often as the real task allows: replay for a heart sound, one or two hearings for a bird's call or a letter sound.
Can I use a recording I found online?
Only on its owner's terms. Freesound (n.d.), for one, sets a licence sound by sound, so read each. Ask whoever handles copyright where you work.
Does EdProcto have a sound recognition template?
Yes: Recognise the sound is in EdProcto's builder now, with a clip, a prompt and two to eight options marked by rule.
References
- Audacity Team. (2026). Loudness normalization. Audacity Manual. https://manual.audacityteam.org/man/loudness_normalization.html
- Barrett, M. J., Lacey, C. S., Sekara, A. E., Linden, E. A., & Gracely, E. J. (2004). Mastering cardiac murmurs: The power of repetition. Chest, 126(2), 470–475. https://doi.org/10.1378/chest.126.2.470
- Creative Commons. (n.d.). Sharing openly, sharing globally. Retrieved October 10, 2026, from https://creativecommons.org/cc-licenses/
- Ehri, L. C., Nunes, S. R., Willows, D. M., Schuster, B. V., Yaghoub-Zadeh, Z., & Shanahan, T. (2001). Phonemic awareness instruction helps children learn to read: Evidence from the National Reading Panel's meta-analysis. Reading Research Quarterly, 36(3), 250–287. https://doi.org/10.1598/RRQ.36.3.2
- Freesound. (n.d.). Frequently asked questions. Retrieved October 10, 2026, from https://freesound.org/help/faq/
- Fritts, L. (n.d.). Musical instrument samples. University of Iowa Electronic Music Studios. Retrieved October 10, 2026, from https://theremin.music.uiowa.edu/MIS.html
- Giordano, B. L., & McAdams, S. (2010). Sound source mechanics and musical timbre perception: Evidence from previous studies. Music Perception, 28(2), 155–168. https://doi.org/10.1525/mp.2010.28.2.155
- Lively, S. E., Logan, J. S., & Pisoni, D. B. (1993). Training Japanese listeners to identify English /r/ and /l/. II: The role of phonetic environment and talker variability in learning new perceptual categories. The Journal of the Acoustical Society of America, 94(3), 1242–1255. https://pmc.ncbi.nlm.nih.gov/articles/PMC3509365/
- Mangione, S., & Nieman, L. Z. (1997). Cardiac auscultatory skills of internal medicine and family practice trainees: A comparison of diagnostic proficiency. JAMA, 278(9), 717–722. https://doi.org/10.1001/jama.1997.03550090041030
- Ogawa, Y., Fukasawa, K., Yoshioka, A., Kumada, N., Takenaka, A., & Ito, T. (2023). Quiz-style online training tool helps to learn birdsong identification and support citizen science. PeerJ, 11, Article e15387. https://doi.org/10.7717/peerj.15387
- Siedenburg, K. (2019). Specifying the perceptual relevance of onset transients for musical instrument identification. The Journal of the Acoustical Society of America, 145(2), 1078–1087. https://doi.org/10.1121/1.5091778
- Trinity College London. (n.d.). Practical exams (face to face). Retrieved October 10, 2026, from https://www.trinitycollege.com/qualifications/music/special-needs/practical
- Wong, S. S. H., Chen, S., & Lim, S. W. H. (2021). Learning melodic musical intervals: To block or to interleave? Psychology of Music, 49(4), 1027–1046. https://doi.org/10.1177/0305735620922595