This paper presents and explores a robust deep learning framework for auscultation analysis. This aims to classify anomalies in respiratory cycles and detect diseases, from respiratory sound recordings. The framework begins with front-end feature extraction that transforms input sound into a spectrogram representation. Then, a back-end deep learning network is used to classify the spectrogram features into categories of respiratory anomaly cycles or diseases. Experiments, conducted over the ICBHI benchmark dataset of respiratory sounds, confirm three main contributions towards respiratory- sound analysis. Firstly, we carry out an extensive exploration of the effect of spectrogram types, spectral-time resolution, overlapping/non-overlapping windows, and data augmentation on final prediction accuracy. This leads us to propose a novel deep learning system, built on the proposed framework, which outperforms current state-of-the-art methods. Finally, we apply a Teacher-Student scheme to achieve a trade-off between model performance and model complexity which holds promise for building real-time applications.

Generative adversarial networks (GAN) have recently been shown to be efficient for speech enhancement. However, most, if not all, existing speech enhancement GANs (SEGAN) make use of a single generator to perform one-stage enhancement mapping. In this work, we propose to use multiple generators that are chained to perform multi-stage enhancement mapping, which gradually refines the noisy input signals in a stage-wise fashion. Furthermore, we study two scenarios: (1) the generators share their parameters and (2) the generators' parameters are independent. The former constrains the generators to learn a common mapping that is iteratively applied at all enhancement stages and results in a small model footprint. On the contrary, the latter allows the generators to flexibly learn different enhancement mappings at different stages of the network at the cost of an increased model size. We demonstrate that the proposed multi-stage enhancement approach outperforms the one-stage SEGAN baseline, where the independent generators lead to more favorable results than the tied generators. The source code is available at http://github.com/pquochuy/idsegan.

Predictive maintenance (PdM) is one of the most powerful maintenance techniques based on the estimation of the remaining useful life (RUL) of machines. Accurately estimating the RUL is crucial to ensure the effectiveness of PdM. However, current methods have limitations in fully exploring condition monitoring data, particularly vibration signals, for RUL estimation. To address these challenges, this research presents a novel Robust Multi-Branch Deep Learning (Robust-MBDL) model. Robust-MBDL stands out by leveraging diverse data sources, including raw vibration signals, time–frequency representations, and multiple feature domains. To achieve this, it adopts a specialized three-branch architecture inspired by efficient network designs. The model seamlessly integrates information from these branches using an advanced attention-based Bi-LSTM network. Furthermore, recognizing the importance of data quality, Robust-MBDL incorporates an unsupervised LSTM-Autoencoder for noise reduction in raw vibration data. This comprehensive approach not only overcomes the limitations of existing methods but also leads to superior performance. Experimental evaluations on benchmark datasets such as XJTU-SY and PRONOSTIA showcase Robust-MBDL’s efficacy, particularly in rotating machine health prognostics. These results underscore its potential for real-world applications, heralding a new era in predictive maintenance practices.

Spectrogram-based representations have grown to dominate the feature space for deep learning audio analysis systems, and are often adopted for speech analysis also. Initially, the primary motivation behind spectrogram-based representations was their ability to present sound as a two-dimensional signal in the time–frequency plane, which not only provides an interpretable physical basis for analysing sound, but also unlocks the use of a range of machine learning techniques such as convolutional neural networks, which had been developed for image processing. A spectrogram is a matrix characterised by the resolution and span of its dimensions, as well as by the representation and scaling of each element. Many possibilities for these three characteristics have been explored by researchers across numerous application areas, with different settings showing affinity for various tasks. This paper reviews the use of spectrogram-based representations and surveys the state-of-the-art to question how front-end feature representation choice allies with back-end classifier architecture for different tasks.