How Ausculta works
Ausculta is an atlas of recorded textbook examples for learning auscultation. This page describes where the examples come from, how they are prepared and described, what the playback tools do, where the marks on the waveform come from, and how the quiz, comparison, site map, cases and daily practice work. The app does not use the microphone, measures nothing and evaluates nothing.
Where the examples come from
The atlas contains 42 sound examples (18 heart, 17 lungs, 7 abdomen) and 4 ECG examples with heart sound. All come from open research databases that allow reuse with attribution; one bowel sound example comes from the freesound.org sound library. The examples are part of the app; nothing is downloaded from the internet.
| Database | What the atlas uses | Licence |
|---|---|---|
| CirCor DigiScope (PhysioNet) | heart sounds from different listening sites (recordings of children) | ODC-By 1.0 |
| EPHNOGRAM (PhysioNet) | heart sound and the ECG and sound examples (adults at rest) | ODC-By 1.0 |
| KAUH (Mendeley Data) | breathing (mostly adults) | CC BY 4.0 |
| SPRSound | breathing (recordings of children) | CC BY 4.0 |
| HF_Lung_V1 | breathing (adults) | CC BY 4.0 |
| HLS-CMDS (UCI) | heart sounds from a clinical teaching manikin | CC BY 4.0 |
| JUST Heart Sound Regurgitation (Mendeley Data) | heart sounds with a murmur (adults) | CC BY 4.0 |
| SMART-Ped (Zenodo) | tracheal breathing (children and adolescents) | CC BY 4.0 |
| Bowel sounds signal (Mansour, figshare) | bowel sounds (healthy adults, sensor next to the navel) | CC BY 4.0 |
| Bowel Sounds Audio Files (Rulli et al., figshare) | bowel sounds (physiological peristalsis) | CC BY 4.0 |
| freesound.org, sound 785273 (logancircle2) | bowel sounds after a meal (contact microphone, not auscultation) | CC BY 4.0 |
In the app every example shows its source and licence; the full citations are under About, authorship & references › Atlas: sources and licences. The database authors are not connected with Ausculta and do not endorse it.
How the examples are chosen
Labels from the authors
Examples are chosen by the labels their authors gave the recordings, for example murmur, wheezes, crackles, stridor or physiological peristalsis. Where a database marks added sounds with time stamps, the excerpt is chosen by them.
A clean excerpt
A 10-12 s excerpt is taken from the recording and inspected as a waveform and spectrogram. Excerpts with speech, crying, rubbing, device tones, strong noise or a saturated (clipped) signal are discarded.
Clear labelling
Examples from a clinical teaching manikin say “manikin” in their title, and the example from the sound library is marked as a contact microphone recording, not auscultation. Where a database does not state the listening site, the app does not state one either.
How the examples are prepared
Every excerpt went through the same processing, so the examples are comfortable to listen to and compare. The processing applies only to the finished recordings from the databases.
Excerpt
Only the chosen 10-12 s section of the original recording is used.
Deep rumble
The DC offset and deep rumble are removed: a 20 Hz high-pass for heart sounds, 80 Hz for breathing and 50 Hz for bowel sounds. The processing is zero-phase, so the sound is not shifted in time.
Resampling
The examples are stored at 8000 samples per second, 16 bit, mono. That is enough for heart, breath and bowel sounds and keeps the app small.
Loudness
All examples are brought to the same average loudness. Peaks are never clipped: where clipping would be a risk, the example stays a little quieter instead.
Fade in and out
The start and end of each example fade in and out smoothly (100 ms), so playback does not start or end with a click.
What the playback tools do
Slower playback, the loop and the filters work only on the finished recording of an example while it plays. The stored file is not changed and nothing is measured. The tools are in the Atlas under the section picker, in Compare two and on the site map.
Slower playback at the same pitch
Speeds 1×, 0.75× and 0.5×. The recording is stretched in time (time stretching in the iOS audio player) and the pitch stays: at 0.5× the example lasts twice as long, but S1, S2, a murmur and wheezes sound just as high as in the original. Plain slow playback would also lower the sound by an octave and change its character.
Loop over a stretch
Drag on the waveform of the recording to select a stretch, for example one cardiac cycle or one breath in, and it repeats over and over; tap the waveform to clear the loop. The loop also works together with slower playback and a filter.
Wide, Bell and Diaphragm filters
A band-pass on the recording as it plays: Wide passes 20 to 2500 Hz, Bell 20 to 220 Hz and Diaphragm 100 to 1200 Hz. The names recall the two sides of a stethoscope chest piece (the bell brings out low sounds, the diaphragm higher ones), but this is not a model of any particular stethoscope. The filter can be switched at any time; the recording stays the same.
Waveform with a cursor
Below the controls is the loudness envelope of the recording with a cursor that shows where you are. It is a display of the stored file, not a measurement.
Where the S1 and S2 marks and the stretches on the waveform come from
In 20 of the 42 examples the waveform carries marks or coloured stretches. All of them come from annotations made by the authors of the source databases. Ausculta only transfers them into the excerpt; it recognises and computes nothing.
Heart: S1 and S2 from CirCor
The CirCor DigiScope database (PhysioNet) provides for every recording a segmentation of the cardiac cycle into the states S1, systole, S2 and diastole. In the 8 heart examples from this database, the centre of every S1 and S2 segment was transferred into the example as a mark (289 marks in all). The examples from EPHNOGRAM, HLS-CMDS and JUST have no marks because these databases provide no segmentation.
Lungs: stretches from SPRSound and HF_Lung_V1
SPRSound provides events with a start and end time for its recordings (wheezes, rhonchi, crackles, stridor), HF_Lung_V1 the phases of inhalation and exhalation and the stretches of adventitious sounds. In the 8 lung examples from these databases the stretches are shown in colour (78 stretches in all); stretches of different kinds may overlap, for example crackles inside an inhalation. The examples from KAUH and SMART-Ped have no stretches; these databases label only the whole recording.
How they are transferred into the excerpt
Every example is an excerpt of a source recording with a known start. The annotation times are shifted by that start and clipped to the length of the example; stretches shorter than 50 ms after clipping are left out. During preparation every excerpt was produced again from the source recording and compared with the file in the app (identical to the last sample), so that the times line up. The marks are rounded to thousandths of a second.
The marks and stretches are displayed annotations of the source databases, not the result of an analysis. Ausculta detects nothing in the recording, counts no heart sounds or intervals, and a missing mark on an example means only that the source provides no annotation for it. A single mark may sit on a quiet heart sound masked by a murmur or in the faded-in start of the example.
How the descriptions are written
Every example has a title and a description of two or three sentences: what you hear in the recording and what textbooks call it, for example where a murmur sits in the cardiac cycle, whether crackles are fine or coarse and in which phase of breathing they occur. The names of lung sounds follow the recommended respiratory sound terminology (CORSA): wheezes, rhonchi, crackles, stridor. The descriptions are based on the label the database authors gave the recording and on textbooks; they do not describe a particular person and give no age or diagnosis from the original record. A property of the database is mentioned only when it explains something, such as the faster beats in recordings of children.
Quiz and comparison
Quiz
The quiz picks 10 examples from the atlas (without Pro only from the free ones), plays them one by one and for each offers up to four names of sound types, one of which is correct; the others come preferably from the same section. After you answer you see the correct type and the textbook description right away, and at the end the number of correct answers. Only your answers are scored.
Compare two
You choose two examples (A and B) and play them side by side, each with its description. It suits pairs that are easy to confuse: fine and coarse crackles, wheezes and rhonchi, normal heart sounds and a murmur.
Heard overview
The app remembers on the device which examples you have already played. Each section shows “Heard X of Y”, and the “Not yet heard” filter shows only the ones you are missing. None of this leaves the device.
Site map and cases
Site map
Every atlas example states its listening site when the source database states it (for example CirCor distinguishes the aortic, pulmonic and tricuspid areas and the apex, SMART-Ped the trachea, Mansour the site next to the navel). The map is a drawn chest, front and back, and abdomen with the sites A, P, E, T, M, lung apex, side of the chest and lung base, back, trachea and next to the navel; a tap opens the examples with that site. A site is never guessed: examples with no stated site (6 of 42) are not on the map, and Erb's point has no example yet because none of the source databases names it. The placement of the points on the drawing is textbook placement, not anatomically exact for any particular person.
Cases
Six short situations (a fellow student with four heart sites, a child at a check-up, a child with noisy breathing, an adult with a cough, an adult at the lower left sternal border, the abdomen before and after a meal) are built only from atlas examples: each station is a site on the map and one example recorded there. The scenario is a textbook text, not a description of a real patient, and the example descriptions give no age or diagnosis from the original record. At each station you listen to the example and choose the type of sound from the options as in the quiz; only your answers are scored. Without Ausculta Pro you walk through the stations with free examples; locked ones are skipped and the app says so.
Spaced repetition
Five boxes
The practice uses the Leitner system: every atlas example has a card in one of boxes 1 to 5. Naming it correctly moves the card up one box, naming it wrongly sends it back to box 1. A card in box 1 is due right away, in box 2 after a day, in box 3 after three days, in box 4 after a week and in box 5 after two weeks (intervals of 0, 1, 3, 7 and 14 days).
Daily session
One session has at most 20 examples: first the ones due for review (lowest box first), then examples you have not yet had in practice, in atlas order. Without Ausculta Pro the free examples are practised, with Pro all of them. A question looks like the quiz: the example plays and you choose the type of sound.
Overview and storage
The overview shows correct answers out of the total by type of sound over the last 30 days; the app keeps the last 500 answers. Cards and answers are stored only on the phone, in the app settings, and are deleted when the app is uninstalled. They describe your answers about the atlas recordings, nothing about your body; nothing is sent.
Why spaced repetition: repeated listening is what works in auscultation (Barrett 2004, see Why auscultation), and spreading the repetitions over days with growing gaps is the principle of the Leitner system.
ECG and heart sound on one time axis
The four examples come from the EPHNOGRAM database, in which the ECG and the heart sound were recorded together in adults at rest. You see both traces one above the other on a shared time axis, and during playback a time marker moves along them, so you can follow how the heart sounds come after the electrical activation. The heart sound has the mains hum (50 Hz and its multiples) removed, is limited to the band up to 1 kHz and is stored at 4000 samples per second. The ECG is in the 0.5-40 Hz band at 250 samples per second. All processing is zero-phase, so the traces are not shifted in time against each other.
Nothing is measured or assessed in the examples: Ausculta shows no intervals and no numbers and does not interpret the ECG trace. The shape of the ECG depends on where the electrodes were placed.
What Ausculta does not do
Ausculta does not use the microphone and picks up no sound or signal from the body. It measures nothing, shows no health values, does not recognise sounds, produces no diagnoses and gives no recommendations. The marks and stretches on the waveform are transferred annotations of the source databases; the filters and slower playback only change how the recording plays. The quiz, the cases and daily practice score only the answer you pick for an atlas example. Ausculta is not a medical device and does not replace an examination by a doctor. Always consult a doctor about health concerns.
What the texts are based on
Sovijärvi ARA, Dalmasso F, Vanderschoot J, et al. Definition of terms for applications of respiratory sounds. Eur Respir Rev. 2000;10(77):597-610 (CORSA).
Bohadana A, Izbicki G, Kraman SS. Fundamentals of lung auscultation. N Engl J Med. 2014;370(8):744-751.
Pasterkamp H, Kraman SS, Wodicka GR. Respiratory sounds: advances beyond the stethoscope. Am J Respir Crit Care Med. 1997;156(3):974-987.
Mangione S, Nieman LZ. Cardiac auscultatory skills of internal medicine and family practice trainees. JAMA. 1997;278(9):717-722.
Barrett MJ, Lacey CS, Sekara AE, Linden EA, Gracely EJ. Mastering cardiac murmurs: the power of repetition. Chest. 2004;126(2):470-475.
Leitner S. So lernt man lernen. Freiburg: Herder; 1972 (the Leitner system of spaced repetition).
The full citations of the databases are in the app under Atlas: sources and licences; the sources for the Why auscultation section are on the home page.
How to cite Ausculta
If you use Ausculta in teaching, a presentation or a publication, please cite it like this:
APA: Šikuta, J. (2026). Ausculta - Auscultation Atlas (Version 2.2) [Mobile app]. App Store. https://apps.apple.com/app/id6819630118
Vancouver: Šikuta J. Ausculta - Auscultation Atlas [mobile application]. Version 2.2. Bratislava: Ján Šikuta; 2026 [cited YYYY Mon DD]. Available from: https://apps.apple.com/app/id6819630118
Please state the version you used (shown on the App Store page). When citing a specific sound, also cite the original database; the source and licence are shown with every example in the app.
Author: Ján Šikuta, MD, PhD · ORCID 0000-0002-4690-0317 · admin@breathune.eu