This repository contains the released respiratory sound database for IEEE BioCAS Respiratory Sound Track Challenges. Please refer to this link (Respiratory Sound Challenge) for more information about the challenges.
Our database is the first open-access respiratory sound database in the pediatric population, aged from 1 month to 18 years old. The respiratory sounds contained in the dataset were recorded at the pediatric respiratory department at Shanghai Children’s Medical Center (SCMC) using Yunting model II Stethoscopes.
The recordings are saved in .wav format with naming rules as follows: Each name is composed of 5 elements separated with underscores, including the patient number, age, gender, the recording location, and the recording number of the participants.
- Patient number (e.g., 65101170)
- Age (e.g., 0.4)
- Gender a. Male (0) b. Female (1)
- Recording location a. left posterior (p1) b. left lateral (p2) c. right posterior (p3) d. right lateral (p4) e. Additional locations (p5-p8) for 2024 and 2025 challenges
- Recording number (e.g., 3246)
The annotations at the record and event level are provided in this database. At the record level, each recording with poor signal quality was annotated as Poor Quality, while the recordings with high signal quality were annotated as Normal, CAS, DAS, or CAS & DAS according to the presence/absence of continuous/discontinuous adventitious respiratory sounds. At the event level, each recording was segmented into multiple respiratory events and annotated as Normal, Rhonchi, Wheeze, Stridor, Coarse Crackle, Fine Crackle, or Wheeze+Crackle.
The annotation information of each recording is saved in .json format with the same filename, which contains the annotation at the record level and event level. The annotation at the record level is Normal, CAS, DAS, CAS & DAS, or Poor Quality. The annotation at event level consists of the start (ms) and the end (ms) of respiratory events, and the corresponding type of respiratory events (Normal, Rhonchi, Wheeze, Stridor, Coarse Crackle, Fine Crackle, Wheeze+Crackle).
An example of annotation file is as follow:
{
"record_annotation": "Normal",
"event_annotation": [
{
"start": "342",
"end": "2515",
"type": "Normal"
},
{
"start": "2557",
"end": "3776",
"type": "Normal"
},
{
"start": "4547",
"end": "5651",
"type": "Normal"
},
{
"start": "6439",
"end": "8065",
"type": "Normal"
},
{
"start": "8363",
"end": "9201",
"type": "Normal"
}
]
}Note: The start and end fields are strings representing time in milliseconds. Some annotation files (especially test sets in later years) may only contain event_annotation without record_annotation.
The SPRSound dataset is organized into different versions corresponding to each challenge year (2022, 2023, 2024, 2025). Each year's challenge provides a new test set while using previously released data for training.
-
Version 2022: Initial release for BioCAS 2022 Challenge (Classification tasks)
- Contains: Training set (1,949 files) + Test set (734 files)
- Location:
BioCAS2022/directory
-
Version 2023: BioCAS 2023 Challenge release (Classification tasks, same as 2022)
- Contains: Test set only (871 files)
- Location:
BioCAS2023/directory - Training: Use BioCAS 2022 dataset
-
Version 2024: BioCAS 2024 Challenge release (Compression & Detection tracks)
- Contains: Test set only (1,704 files)
- Location:
BioCAS2024/directory - Training: Use BioCAS 2022 and 2023 datasets
-
Version 2025: BioCAS 2025 Challenge release (Compression & Detection tracks)
- Contains: Test set only (1,309 files)
- Location:
BioCAS2025/directory - Training: Use BioCAS 2022, 2023, and 2024 datasets
Each challenge year has its own directory with a detailed README.md file:
SPRSound/
├── BioCAS2022/ # BioCAS 2022 Challenge dataset
│ ├── README.md # Detailed documentation
│ ├── train2022_wav/ # Training WAV files (1,949 files)
│ ├── train2022_json/ # Training JSON annotations (1,949 files)
│ ├── test2022_wav/ # Test WAV files (734 files)
│ └── test2022_json/ # Test JSON annotations (734 files)
│ ├── inter_test_json/ # Inter-subject test set (355 files)
│ └── intra_test_json/ # Intra-subject test set (379 files)
├── BioCAS2023/ # BioCAS 2023 Challenge test set
│ ├── README.md # Detailed documentation
│ ├── test2023_wav/ # Test WAV files (871 files)
│ └── test2023_json/ # Test JSON annotations (871 files)
├── BioCAS2024/ # BioCAS 2024 Challenge test set
│ ├── README.md # Detailed documentation
│ ├── test2024_wav/ # Test WAV files (1,704 files)
│ └── test2024_json/ # Test JSON annotations (1,704 files)
├── BioCAS2025/ # BioCAS 2025 Challenge test set
│ ├── README.md # Detailed documentation
│ ├── test2025_wav/ # Test WAV files (1,309 files)
│ └── test2025_json/ # Test JSON annotations (1,309 files)
├── Patient Summary/ # Patient summary CSV files
├── example/ # Example files
└── scripts/ # Utility scripts
Important: Each year's challenge provides only the test set. For training, please use the training data from the BioCAS 2022 release. Each version directory contains a detailed README.md file with specific information about that year's dataset, including file naming rules, annotation formats, and challenge tasks.
Task 1-1 is a binary class classification challenge (Normal and Adventitious).
Task 1-2 is a multiclass classification challenge (Normal (N), Rhonchi (R), Wheeze (W), Stridor (S), Coarse Crackle (CC), Fine Crackle (FC), Wheeze & Crackle (WC)).
Task 2-1 is a ternary class classification challenge (Normal, Adventitious, and Poor Quality records).
Task 2-2 is a multiclass classification challenge (Normal (N), CAS (C), DAS (D), CAS & DAS (CD), or Poor Quality (PQ) records).
This track deals with respiratory recordings compression using compressive sensing-based compression methods.
This track deals with the detection of onsets and offsets in addition to the assignment of event labels of respiratory events in respiratory recordings using sound event detection methods.
This track deals with respiratory recordings compression using compressive sensing-based compression methods.
This track deals with the detection of onsets and offsets in addition to the assignment of event labels of respiratory events in respiratory recordings using sound event detection methods.
For use in publications and presentations, please cite this data collection as follows:
[Data] Q. Zhang, et al. “SPRSound: Open-Source SJTU Paediatric Respiratory Sound Database”, IEEE Transactions on Biomedical Circuits and Systems (TBioCAS), vol. 16, no. 5, pp. 867-881, Oct, 2022.
[Competition] Q. Zhang, et al. “Meta: Data Compression and Event Detection Grand Challenge 2024 With SPRSound Dataset”, IEEE Data Descriptions (DD), vol. 1, no. 1, pp. 1-8, Dec, 2024
[Competition] Q. Zhang, et al. “Grand Challenge on Respiratory Sound Classification for SPRSound Dataset II”, IEEE Biomedical Circuits and Systems Conference (BioCAS), pp. 1-5, Oct, 2023.
[Competition] Q. Zhang, et al. “Grand Challenge on Respiratory Sound Classification”, IEEE Biomedical Circuits and Systems Conference (BioCAS), 2022, pp. 1-5.