Prof. Hanna Bogucka, head of the Department of Wireless Communications at the Poznań University of Technology, discusses unnecessary inhibitions, the usefulness of microphones, and the links between people and technology.

Microphone array with minimum variance (MVDR) beamformer is a commonly used method for ambient noise suppression. Unfortunately, the performance of the MVDR beamformer is poor in a real reverberant room due to multipath wave propagation. To overcome this problem, we propose three improvements. Firstly, we propose end-fire microphone array that has been shown to have a better directivity index than the corresponding broadside microphone array. Secondly, we propose the use of unidirectional microphones instead of omnidirectional ones. Thirdly, we propose an adaptation of its adaptive algorithm during the pause of speech, which improves its robustness against the room reverberation and deviation from the optimal receiving direction. The performance of the proposed microphone array was theoretically analyzed using a diffuse noise model. Simulation analysis was performed for combined diffuse and coherent noise using the image model of the reverberant room. Real room tests were conducted using a four-microphone array placed in a small office room. The theoretical analysis and the real room tests showed that the proposed solution considerably improves speech quality.

As a consequence of recent implementations of EU Directives related to noise protection more and more students of various AGH-UST programs are introduced to the basics of acoustic measurements. Students at various levels of theoretical background in the field of acoustic measurements are offered practical training in measurements using digital sound analyzers. The situation would be optimal if each student could have a device at his/her own disposal. Unfortunately, such a situation is not possible at the moment because of various reasons.

With the above problem in mind, a dedicated software package has been developed, implemented in the LabVIEW environment, which allows detailed studies of problems related to the acoustic signal measurement using sound level meters, as well as tasks in spectral analysis (1/1 and 1/3 band filters) and narrow-band (FFT) analysis. With such organization during the introductory laboratory classes each student is offered a direct individual contact with a virtual device that is properly pre-programmed for realization of a well-constructed learning process. It definitely facilitates understanding of the essence of acoustic signal measurements and provides a good basis for further laboratory work carried out as a team-activity.

This paper develops an automatic method to calculate the macrotexture depth of pavement roads, using the tire/road noise data collected by the two directional microphones mounted underneath a moving test vehicle. The directional microphones collect valid tire/road noise signal at the travel speed of 10–110 km/h, and the sampling frequency is 50 kHz. The tire/road noise signal carries significant amount of road surface information, such as macrotexture depth. Using bandpass filter, principal component analysis, speed effect elimination, Gaussian mixture model, and reversible jump Markov Chain Monte Carlo, the macrotexture depth of pavement roads can be calculated from the tire/road noise data, automatically and efficiently. Compared to the macrotexture depth results by the sand-patch method and laser profiler, the acoustic method has been successfully demonstrated in engineering applications for the accurate results of macrotexture depth with excellent repeatability, at the test vehicle’s travel speed of 10-110 km/h.

Acoustic source localization using distributed microphone array is a challenging task due to the influences of noise and reverberation. In this paper, acoustic source localization using kernel-based extreme learning machine in distributed microphone array is proposed. Specifically, the space of interest is divided into some labeled positions, and the candidate generalized cross correlation function in each node is treated as the feature mapped into the hidden nodes of extreme learning machine. During the training phase, by the implementation of kernel function, the output weights of the classifier are calculated and do not need to be tuned. After the kernel-based extreme learning machine (K-ELM) is well trained, the measured generalized cross correlation data are fed into the K-ELM classifier, and the output is the estimated acoustic source position. The proposed method needs less human intervention for both training and testing and it does not need to calibrate the node in advance. Simulation and real-world experimental results reveal that the proposed method has extremely fast training and testing speeds, and can obtain better localization performance than steered response power, K-nearest neighbor, and support vector machine methods.

Research work on the design of robust multimodal speech recognition systems making use of acoustic and visual cues, extracted using the relatively noise robust alternate speech sensors is gaining interest in recent times among the speech processing research fraternity. The primary objective of this work is to study the exclusive influence of Lombard effect on the automatic recognition of the confusable syllabic consonant-vowel units of Hindi language, as a step towards building robust multimodal ASR systems in adverse environments in the context of Indian languages which are syllabic in nature. The dataset for this work comprises the confusable 145 consonant-vowel (CV) syllabic units of Hindi language recorded simultaneously using three modalities that capture the acoustic and visual speech cues, namely normal acoustic microphone (NM), throat microphone (TM) and a camera that captures the associated lip movements. The Lombard effect is induced by feeding crowd noise into the speaker’s headphone while recording. Convolutional Neural Network (CNN) models are built to categorise the CV units based on their place of articulation (POA), manner of articulation (MOA), and vowels (under clean and Lombard conditions). For validation purpose, corresponding Hidden Markov Models (HMM) are also built and tested. Unimodal Automatic Speech Recognition (ASR) systems built using each of the three speech cues from Lombard speech show a loss in recognition of MOA and vowels while POA gets a boost in all the systems due to Lombard effect. Combining the three complimentary speech cues to build bimodal and trimodal ASR systems shows that the recognition loss due to Lombard effect for MOA and vowels reduces compared to the unimodal systems, while the POA recognition is still better due to Lombard effect. A bimodal system is proposed using only alternate acoustic and visual cues which gives a better discrimination of the place and manner of articulation than even standard ASR system. Among the multimodal ASR systems studied, the proposed trimodal system based on Lombard speech gives the best recognition accuracy of 98%, 95%, and 76% for the vowels, MOA and POA, respectively, with an average improvement of 36% over the unimodal ASR systems and 9% improvement over the bimodal ASR systems.

The acoustic vector sensor (AVS) is used to measure the acoustic intensity, which gives the direction-ofarrival (DOA) of an acoustic source. However, while estimating the DOA from the measured acoustic intensity the finite microphone separation (d) in a practical AVS causes angular bias. Also, in the presence of noise there exists a trade off between the bias (strictly increasing function of d) and variance (strictly decreasing function of d) of the DOA estimate. In this paper, we propose a novel method for mitigating the angular bias caused due to finite microphone separation in an AVS. We have reduced the variance by increasing the microphone separation and then removed the bias with the proposed bias model. Our approach employs the finite element method (FEM) and curves fitting to model the angular bias in terms of microphone separations and frequency of a narrowband signal. Further, the bias correction algorithm based on the intensity spectrum has been proposed to improve the DOA estimation accuracy of a broadband signal. Simulation results demonstrate that the proposed bias correction scheme significantly reduces the angular bias and improves the root mean square angular error (RMSAE) in the presence of noise. Experiments have been performed in an acoustic full anechoic room to corroborate the effect of microphone separation on DOA estimation and the efficacy of the bias correction method.