Clinically Informed Acoustic-Aware Attribute Modeling for Respiratory Sound Classification
Authors: Xinhong Li, Li Xiao, Ziyang Chen, Yuhong Yang, Weiping Tu
Organizations: School of Computer Science, Wuhan University, Wuhan, China · Hubei Key Laboratory of Multimedia and Network Communication Engineering, Wuhan University, Wuhan, China
Automated respiratory sound classification (RSC) aims to identify normal and adventitious respiratory sounds, such as crackles and wheezes. Most existing methods formulate RSC as a flat four-class classification problem over norma, crackle, wheeze and both. However, unlike conventional classification tasks with semantically independent and mutually exclusive categories, the commonly used four-class RSC labels have a compositional structure and can be represented by two clinically meaningful sound attributes: crackle presence and wheeze presence. Existing flat classifiers do not explicitly exploit this structure, which may increase the learning burden by treating the four labels as independent targets. Meanwhile, crackles and wheezes exhibit distinct temporal characteristics, with sparse transient crackle events and sustained continuous wheeze patterns, making a single temporal aggregation strategy less suitable for modeling different adventitious sound evidence. To address these limitations, we propose an Acoustic-Aware Asymmetric Framework (AAAF) for RSC. AAAF reformulates four-class recognition as two sound-attribute detection tasks and recovers the final category through logical composition during inference. Built upon a pretrained BEATs encoder, AAAF employs branch-specific multi-layer fusion and asymmetric temporal aggregation to capture localized crackle evidence and sustained wheeze patterns. Auxiliary normal--abnormal supervision and a batch-adaptive dynamic loss are further introduced to improve global discrimination and robustness under imbalanced labels. Experiments on ICBHI and SPRSound demonstrate that AAAF improves overall RSC performance and generalizes across datasets, achieving a state-of-the-art four-class Score of 65.30% on ICBHI.
Figures & tables
Fig. 1: Motivation of AAAF: flat four-class RSC versus crackle–wheeze attribute modeling with acoustic-aware temporal aggregation.
Fig. 2: Overview of the proposed AAAF. Hierarchical BEATs representations are fused and fed into two attribute-specific branches for crackle and wheeze presence detection. Temporal Top- K pooling emphasizes localized crackle evidence, while average pooling with channel attention captures sustained wheeze patterns. The final four-class prediction is obtained by logical composition, with auxiliary normal–abnormal supervision and batch-adaptive dynamic loss for joint optimization.
Method
Architecture
Pretrain
Venue
Sp (%)
Se (%)
Score (%)
4-class eval.
Patch-Mix [ 15 ]
AST
IN + AS
INTERSPEECH’23
81.66
43.07
62.37
SG-SCL [ 6 ]
AST
IN + AS
ICASSP’24
79.87
43.55
61.71
RepAugment [ 14 ]
AST
IN + AS
EMBC’24
82.47
40.55
61.51
LungAdapter [ 5 ]
AST
IN + AS
INTERSPEECH’24
80.43
44.37
62.40
BTS [ 21 ]
CLAP
LA
INTERSPEECH’24
81.40
45.67
63.54
AMG-SCL [ 22 ]
AST
IN + AS
JBHI’25
76.18
47.91
62.04
TABLE I: Performance comparison on the ICBHI 2017 benchmark. Bold and underline indicate the best and second-best results, respectively. IN, AS, and LA denote ImageNet [ 18 ] , AudioSet [ 19 ] , and LAION-Audio-630K [ 20 ] pretraining.
Method
Architecture
Pretrain
4-class eval.
2-class eval.
Sp (%)
Se (%)
Score (%)
Sp (%)
Se (%)
Score (%)
ResNet-50 †
CNN
IN
91.33
82.81
87.07
89.72
90.07
89.90
EfficientNet-B0 †
CNN
IN
91.68
83.57
87.63
90.30
91.15
90.72
AST †
AST
IN + AS
90.85
84.65
87.75
89.80
92.57
91.19
BEATs †
BEATs
AS
91.39
85.36
88.38
90.06
93.99
92.02
SCL [ 4 ]
CNN
AS
90.49
80.69
85.59
–
–
–
TABLE II: External validation results on the SPRSound dataset. † denotes results reproduced by us under the same evaluation protocol. Pretraining abbreviations follow Table I .
TABLE IV: Comparison of BEATs layer selection strategies for hierarchical feature extraction.
K
Sp (%)
Se (%)
Score (%)
1
82.13 ± 4.03
46.29 ± 3.77
64.21 ± 0.84
3
79.38 ± 2.68
50.09 ± 1.74
64.74 ± 0.64
5
80.25 ± 3.57
50.35 ± 3.15
65.30 ± 0.50
7
79.54 ± 7.57
48.84 ± 6.35
64.19 ± 1.00
9
84.23 ± 6.32
45.18 ± 7.19
64.70 ± 0.94
11
80.11 ± 5.15
49.02 ± 4.65
64.57 ± 0.50
TABLE V: Impact of Top- K pooling in the crackle branch.
Fig. 3: Frame-wise branch responses on a Both sample before temporal pooling. The crackle branch produces sparse peak responses around localized transient regions, with selected Top- K frames marked by dots, while the wheeze branch shows broader sustained activations over longer intervals.
Backbone
Method
Sp (%)
Se (%)
Score(%)
ResNet-50
CE
76.53 ± 4.56
32.01 ± 2.95
54.24 ± 1.00
AAAF
79.11 ± 2.02
31.47 ± 2.71
55.29 ± 0.71
Eff.-B0
CE
77.63 ± 3.94
37.72 ± 3.19
57.68 ± 0.79
AAAF
86.92 ± 0.87
31.05 ± 1.46
58.98 ± 0.66
AST
CE
73.12 ± 3.38
45.91 ± 2.19
59.52 ± 0.84
AAAF
75.27 ± 4.46
47.03 ± 5.19
61.15 ± 0.51
TABLE VI: Backbone generalization on ICBHI four-class evaluation.
Respiratory sound classification distinguishes four categories: Normal, Crackle, Wheeze, and Both (crackle and wheeze). Existing knowledge distillation transfers four-class predictions without explicitly exploiting shared respiratory events, potentially limiting the transfer of event-specific knowledge across individual abnormal classes and the Both class. We propose PAC-KD, a pathology-aware compositional knowledge distillation framework that learns crackle and wheeze presence through two dedicated specialists. Their complementary knowledge is transferred to a unified four-class student, enabling it to effectively capture pathology-specific attributes while preserving class-level discrimination. Furthermore, stable conflict calibration adaptively regulates specialist guidance according to its consistency with classification supervision, mitigating potential supervision conflicts. Experiments on ICBHI and SPRSound achieve Scores of 65.28% and 89.54%, respectively, outperforming strong knowledge distillation baselines on both datasets.
Li Xiao, Weiping Tu, Youqiang Zheng +1
School of Computer Science, Wuhan University, Wuhan 430072, China
Recent respiratory sound classification (RSC) studies largely rely on CLS-token driven self-attention architectures such as the Audio Spectrogram Transformer (AST). While effective at modeling global context, recent analyses suggest a low-pass filtering behavior that may reduce sensitivity to localized abnormal patterns. In this work, we investigate State Space Models (SSMs) as an alternative backbone for RSC. Using the Distilled Audio State Space model, we analyze intermediate representations through spectral response curves and observe stronger preservation of mid-to-high spatial-frequency components. Based on these observations, we introduce spectral-aware layer regularization using Gaussian convolution applied to selected layers. We further propose Dual-Axis Patch-Mix contrastive learning tailored to SSM-based audio models for robust representation learning. Experiments on the ICBHI benchmark show that our approach achieves 64.48% score, outperforming the AST baseline by 5%. Code is available at https://github.com/RSC-Toolkit/Lung-SRAD.
Hemansh Shridhar, Miika Toikkanen, June-Woo Kim
RSC LAB, MODULABS, Republic of Korea · Department of Electronic Engineering, Wonkwang University, Republic of Korea · AI Convergence Research Institute, Wonkwang University, Republic of Korea
Respiratory auscultation provides a low-cost means of assessing pulmonary function, yet transient crackles and sustained wheezes can be difficult to recognize in noisy clinical recordings. Conventional deep learning classifiers often operate as black boxes, offering limited insight into the temporal acoustic evidence supporting their predictions. We present RespFocus, a weakly supervised framework that uses multiple instance learning (MIL) to learn inspectable temporal evidence from cycle-level labels and integrate it into classification. We also introduce PedResp2026, a multicenter pediatric respiratory sound cohort collected in routine clinical environments with adjudicated cycle-level annotations. In classification experiments, RespFocus achieves four-class Scores of 69.55% on public ICBHI 2017 and 72.91% on PedResp2026. Evidence diagnostics and matched temporal interventions support the functional contribution of the learned evidence to model predictions. Together, these findings highlight RespFocus's potential as a tool for automated respiratory sound diagnosis.
Yi Tian, Xiaohuan Chen, Yufei tang +5
Fuyao University of Science and Technology (fyust.edu.cn) · Xiamen University (xmu.edu.cn)