HLS-CMDS: Heart and Lung Sounds Dataset Recorded from a Clinical Manikin using Digital Stethoscope

** Accepted in IEEE Data Descriptions Journal **
This dataset contains 535 recordings of heart and lung sounds captured using a digital stethoscope from a clinical manikin, including both individual and mixed recordings of heart and lung sounds; It includes 50 heart sounds, 50 lung sounds, and 145 mixed sounds. For each mixed sound, the corresponding source heart sound (145 recordings) and source lung sound (145 recordings) were also recorded. It includes recordings from different anatomical chest locations, with normal and abnormal sounds. Each recording has been filtered to highlight specific sound types, making it valuable for artificial intelligence (AI) research and applications in automated cardiopulmonary disease detection, sound classification, and deep learning algorithms related to audio signal processing. If you use this dataset in your research, please cite the following paper:

Y. Torabi, S. Shirani and J. P. Reilly, "Descriptor: Heart and Lung Sounds Dataset Recorded from a Clinical Manikin using Digital Stethoscope (HLS-CMDS)," in IEEE Data Descriptions, https://doi.org/10.1109/IEEEDATA.2025.3566012 .

Data Type: Audio files (.wav), Comma Separated Values (.CSV)

Each category is accompanied by a corresponding CSV file that provides metadata for the respective audio files. The CSV files (HS.csv, LS.csv, and Mix.csv) contain metadata about the corresponding audio files, including the file name, gender, heart and lung sound type, and the anatomical location where we recorded the sound.

Sound Types: Normal Heart, Late Diastolic Murmur, Mid Systolic Murmur, Late Systolic Murmur, Atrial Fibrillation, Fourth Heart Sound, Early Systolic Murmur, Third Heart Sound, Tachycardia, Atrioventricular Block, Normal Lung, Wheezing, Fine Crackles, Rhonchi, Pleural Rub, and Coarse Crackles.

Auscultation Landmarks: Right Upper Sternal Border, Left Upper Sternal Border, Lower Left Sternal Border, Right Costal Margin, Left Costal Margin, Apex, Right Upper Anterior, Left Upper Anterior, Right Mid Anterior, Left Mid Anterior, Right Lower Anterior, and Left Lower Anterior.

Applications: AI-based cardiopulmonary disease detection, unsupervised sound separation techniques, and deep learning for audio signal processing.

**已被《IEEE数据描述期刊》（IEEE Data Descriptions Journal）收录**
本数据集包含535条心音与肺音录音，均通过电子听诊器（digital stethoscope）从临床模拟人（clinical manikin）采集，涵盖心音、肺音的单类录音与混合录音两类。数据集包含50条心音录音、50条肺音录音及145条混合录音。针对每一条混合录音，同时采集了对应的原始心音录音（145条）与原始肺音录音（145条）。录音覆盖胸部不同解剖位置，包含正常与异常心/肺音。每条录音均经过滤波处理以突出特定声音类型，可用于人工智能（AI）研究与自动化心肺疾病（cardiopulmonary disease）检测、声音分类以及音频信号处理（audio signal processing）相关深度学习（deep learning）算法等领域的应用。

若您在研究中使用本数据集，请引用以下论文：
Y. Torabi、S. Shirani与J. P. Reilly，《数据描述：基于电子听诊器从临床模拟人采集的心肺音数据集（HLS-CMDS）》，发表于《IEEE数据描述》，https://doi.org/10.1109/IEEEDATA.2025.3566012。

数据类型：音频文件（.wav）、逗号分隔值文件（.CSV）。
每个类别均配有对应的CSV文件，用于提供对应音频文件的元数据。本次发布的HS.csv、LS.csv及Mix.csv三类CSV文件，包含对应音频文件的元数据，具体包括文件名、受试者性别、心/肺音类型以及录音采集的胸部解剖位置。

声音类型：正常心音、舒张晚期杂音（Late Diastolic Murmur）、收缩中期杂音（Mid Systolic Murmur）、收缩晚期杂音（Late Systolic Murmur）、心房颤动（Atrial Fibrillation）、第四心音（Fourth Heart Sound）、收缩早期杂音（Early Systolic Murmur）、第三心音（Third Heart Sound）、心动过速（Tachycardia）、房室传导阻滞（Atrioventricular Block）、正常肺音（Normal Lung）、哮鸣音（Wheezing）、细湿啰音（Fine Crackles）、干啰音（Rhonchi）、胸膜摩擦音（Pleural Rub）、粗湿啰音（Coarse Crackles）。

听诊标志位置：胸骨右缘上区（Right Upper Sternal Border）、胸骨左缘上区（Left Upper Sternal Border）、胸骨左缘下区（Lower Left Sternal Border）、右肋缘（Right Costal Margin）、左肋缘（Left Costal Margin）、心尖部（Apex）、右上前胸区（Right Upper Anterior）、左上前胸区（Left Upper Anterior）、右中前胸区（Right Mid Anterior）、左中前胸区（Left Mid Anterior）、右下前胸区（Right Lower Anterior）、左下前胸区（Left Lower Anterior）。

应用场景：基于AI的心肺疾病检测、无监督声音分离技术以及音频信号处理相关深度学习研究。

搜集汇总

背景与挑战

背景概述

HLS-CMDS数据集包含535个心脏和肺部声音的录音，分为单独和混合声音，涵盖多种正常和异常声音类型。数据集提供了详细的元数据，适用于人工智能研究和自动化心肺疾病检测等应用。