Table 4.
Open source Datasets.
| Reference | Title | Description | Provider | Suitable for Respiratory Disease Classification | Suitable for Cough Detection |
|---|---|---|---|---|---|
| [92] | COUGHVID | Over 25,000 crowdsourced audio recordings: Cough—a wide range of participant ages, genders, geographic locations, and COVID-19 statuses | Embedded Systems Laboratory (ESL), EPFL, Lausanne, Switzerland | X | ✓ |
| [49] | COVID-19 Sounds | 53,449 audio recordings, over 552 h in total: 3 Cough, 3–5 Breathing, 3 Speech of users reading a specific sentence | University of Cambridge | X | ✓ |
| [39] | Coswara | 2747 audio recordings: Breathing, Coughing, Talking—Crowdsourced dataset (not clinically validated) | Indian Institute of Science (IISc), Bangalore | ✓ | ✓ |
| [69] | Respiratory Sound Database (RSDB) | 920 audio recordings: Crackles or/and Wheezes - Digital stethoscopes and microphones, each recording is expertly annotated | Department of Informatics Engineering, University of Coimbra, Portugal and School of Medicine, Aristotle University of Thessaloniki, Greece | ✓ | X |
| [40] | Corp | 168 h of 9969 audio recordings: Cough—42 different patients with respiratory diseases | MARI Lab, Tongji university | X | ✓ |
| [37] | Virufy | Combination of Coswara & COUGHVID audio recordings: Cough—COVID-19 positive/negative | The Covid Detection Foundation (California nonprofit corporation) | X | ✓ |
| [93] | COVID-19 and Pulmonary Abnormalities | 1734 COVID-19 spectrogram images of respiratory sounds: 795 Crackles, 322 Wheezes, 1143 Normal. | Indian Institute of Science, PES University, M S Ramaiah Institute of Technology, Concordia University | ✓ | X |
| [94] | Tos-COVID | Audio recordings: Cough | Gov. of Buenos Aires city | X | ✓ |
| [95] | SPRSound: Open-Source SJTU Paediatric Respiratory Sound Database | 2683 audio recordings and 9089 audio events: Respiratory Symptoms/Sounds—292 participants. | Shanghai Jiao Tong University and Shanghai Children’s Medical Center (SCMC) | ✓ | X |
| [96] | HF_Lung | Audio recordings: Lung Sounds/Symptoms—Used for developing automated inhalation, exhalation, and adventitious sound detection algorithms | Taiwan Smart Emergency and Critical Care (TSECC) and Taiwan Society of Emergency and Critical Care Medicine (TSECCM) | ✓ | X |
| [97] | ESC-50 | 2000 audio recordings: Environmental, Various, Cough—Labeled collection suitable for benchmarking methods of sound classification | Warsaw University of Technology, Warsaw, Poland | ∼ | ✓ |
| [98] | AudioSet | 2,084,320 10-s audio recordings: Environmental, Various, Cough, Respiratory Symptoms—Expanding ontology, 632 human-labeled audio event classes, drawn from YouTube videos | Sound and Video Understanding teams, Google LLC | ∼ | ✓ |