Explore the vast range of titles available.

Machinery health monitoring techniques provide valuable insights into the performance and condition of machines. Acoustic sensor-based monitoring has emerged as a significant area of interest for the industry due to its ability to accurately capture fault signatures, thereby improving the detection accuracies of anomalies or deviations from regular operations. However, the collected sensor signals typically contain a complex mixture of sounds that relate to multiple fault conditions, environmental noise, and other unwanted sounds from the surroundings. Identifying the specific root causes of failures is a challenge in modeling without knowledge of the unique characteristics of failure conditions. This can ultimately degrade the model’s performance or yield inaccurate failure estimations in condition monitoring, which is a consistent concern in the industry. Therefore, this study proposes a novel framework that enhances the accuracy of machinery fault diagnosis using audio source separation of a complex mixture of sound signals. The proposed approach employs a Deep Extractor for Music Source Separation (DEMUCS), a state-of-the-art music source separation approach consisting of an encoder-decoder architecture that uses bidirectional long short-term memory (LSTM) for industrial machine sound separation and enhancement. The proposed methodology comprises two steps. In the first step, the fault sound isolation and recovering individual fault sounds from a complex mixture of sound signals are enabled using DEMUCS. In the second step, the isolated fault sounds are fed through a 1D-convolutional neural network (1D-CNN) classifier for adequate classification. A machine fault simulator by Spectra Quest equipped with a condenser mic was employed to evaluate the proposed DEMUCS-CNN methodology for identifying multiple faults. The effectiveness of the DEMUCS-CNN method was also compared to the traditional approach of blind source separation (BSS). The outcomes of the comparison indicated that the suggested approach of fault isolation by DEMUCS led to enhanced fault classification accuracy, making it a more effective approach compared to conventional BSS.

The Demucs-Denoiser model has been recently shown to achieve a high level of performance for online speech enhancement, but assumes that only one speech source is present in the fed mixture. In real-life multiple-speech-source scenarios, it is not certain which speech source will be enhanced. To correct this issue, two target selection strategies for the Demucs-Denoiser model are proposed and evaluated: (1) an embedding-based strategy, using a codified sample of the target speech, and (2) a location-based strategy, using a beamforming-based prefilter to select the target that is in front of a two-microphone array. In this work, it is shown that while both strategies improve the performance of the Demucs-Denoiser model when one or more speech interferences are present, they both have their pros and cons. Specifically, the beamforming-based strategy achieves overall a better performance (increasing the output SIR between 5 and 10 dB) compared to the embedding-based strategy (which only increases the output SIR by 2 dB and only in low-input-SIR scenarios). However, the beamforming-based strategy is sensitive against the location variation of the target speech source (decreasing the output SIR by 10 dB if the target speech source is located only 0.1 m from its expected position), which the embedding-based strategy does not suffers from.

Audio effects are an ubiquitous tool in music production due to the interesting ways in which they can shape the sound of music. Guitar effects, the subset of all audio effects focusing on guitar signals, are commonly used in popular music to shape the guitar sound to fit specific genres or to create more variety within musical compositions. Automatic extraction of guitar effects and their parameter settings, with the aim to copy a target guitar sound, has been previously investigated, where artificial neural networks first determine the effect class of a reference signal and subsequently the parameter settings. These approaches require a corresponding guitar effect implementation to be available. In general, for very close sound matching, additional research regarding effect implementations is necessary. In this work, we present a different approach to circumvent these issues. We propose blind extraction of guitar effects through a combination of blind system inversion and neural guitar effect modeling. That way, an immediately usable, blind copy of the target guitar effect is obtained. The proposed method is tested with the phaser, softclipping and slapback delay effect. Listening tests with eight subjects indicate excellent quality of the blind copies, i.e., little to no difference to the reference guitar effect.