Article
Multidisciplinary Sciences
Wael Alosaimi, Md Tarique Jamal Ansari, Abdullah Alharbi, Hashem Alyami, Adil Hussain Seh, Abhishek Kumar Pandey, Alka Agrawal, Raees Ahmad Khan
Summary: AAL is an interdisciplinary field aimed at improving the lives of the elderly through technology. AAL systems require high-performance functionality to ensure interoperability, accessibility, security, and consistency. Standardization, continuity, and system development assistance are urgently needed to meet the growing demands.
Article
Multidisciplinary Sciences
Wiktoria Wilkowska, Julia Offermann, Susanna Spinsante, Angelica Poli, Martina Ziefle
Summary: This study investigated the technology acceptance and privacy perceptions of sensor-based applications implemented in private environments. The results showed that participants recognized the benefits of these technologies but had limited intention to use them. They highly valued their privacy and considered the collection of data in intimate spaces as problematic.
Article
Computer Science, Information Systems
Jie Wan, MingSong Li, Michael J. OGrady, Xiang Gu, Munassar A. A. H. Alawlaqi, Gregory M. P. OHare
Summary: Real-time activity recognition is essential for smart homes, but is currently dominated by the research community using machine learning and AI techniques. Research mainly relies on pre-segmented data, which may not be sufficient for assistive paradigms dependent on smart technologies.
IEEE TRANSACTIONS ON EMERGING TOPICS IN COMPUTING
(2021)
Review
Information Science & Library Science
Gloria Puliga, Akhatjon Nasullaev, Flavio Bono, Eugenio Gutierrez, Fernanda Strozzi
Summary: The authors analyzed the impact of European funding research programmes on Ambient Assisted Living technology through a systematic literature review. They found that the funding programs led to increased scientific production and identified three main research directions: activity and vital sign recognition, human-computer interaction, and technology acceptance.
INFORMATION TECHNOLOGY & PEOPLE
(2021)
Article
Chemistry, Analytical
Caetano Mazzoni Ranieri, Scott MacLeod, Mauro Dragone, Patricia Amancio Vargas, Roseli Aparecida Francelin Romero
Summary: Global demographic projections showing an aging population have driven research on Ambient Assisted Living, focusing on smart homes and social robots. This paper addresses the recognition of heterogeneous daily living activities in home environments using data from videos, wearable IMUs, and ambient sensors. The study introduces a Deep Learning framework for multimodal activity recognition in collaboration with the HWU-USP dataset, demonstrating improved accuracy results through the integration of ambient sensor data.
Article
Nanoscience & Nanotechnology
Boling Lan, Tao Yang, Guo Tian, Yong Ao, Long Jin, Da Xiong, Shenglong Wang, Hongrui Zhang, Lin Deng, Yue Sun, Jieling Zhang, Weili Deng, Weiqing Yang
Summary: This article introduces a cochlear-inspired multichannel piezoelectric acoustic sensor (MAS) based on gradient PVDF piezoelectric nanofibers for broadband voice recognition. Compared with traditional piezoelectric ceramic or piezoelectric fiber acoustic sensors, MAS has a wider frequency response range and a stronger piezoelectric output. Furthermore, this sensor can be used for high-fidelity music recording and human voice recognition, with a classification accuracy rate of up to 100% when combined with deep learning. This research has universal significance for the development of intelligent bioelectronics.
ACS APPLIED MATERIALS & INTERFACES
(2023)
Review
Computer Science, Information Systems
Mohamed-Amine Choukou, Taylor Shortly, Nicole Leclerc, Derek Freier, Genevieve Lessard, Louise Demers, Claudine Auger
Summary: This study aims to explore the acceptance evaluation of AALT technologies in rehabilitation contexts. A total of 51 AALTs dedicated to various rehabilitation contexts were identified, with focus on monitoring OAs' activities and environmental changes. Although older adults intend to use and perceive the usefulness of AALTs, they still have concerns and hesitation about adopting the technology, suggesting the need for more comprehensive and standardized methodologies in assessing AALT acceptance.
INTERNATIONAL JOURNAL OF MEDICAL INFORMATICS
(2021)
Article
Chemistry, Analytical
Bruna Maria Vittoria Guerra, Micaela Schmid, Giorgio Beltrami, Stefano Ramat
Summary: Human Action Recognition (HAR) is an evolving field that has impacts on various domains, including Ambient Assisted Living (AAL). This study proposes a monitoring system that detects dangerous situations by classifying human postures using Artificial Intelligence (AI) solutions. The analysis shows that the LSTM approach has better suitability and achieves higher performance compared to the MLP approach.
Article
Automation & Control Systems
Lucia Cascone, Michele Nappi, Fabio Narducci, Ignazio Passero
Summary: The article discusses the development of VPepper, a virtual replica of the Pepper robot, and its interaction with smart objects in a smart home environment. Through the use of digital twin, machine learning procedures can be seamlessly transferred between the virtual replica and the physical robot. The practical application shows promising opportunities for simulation accuracy and machine learning instruments in real settings.
IEEE TRANSACTIONS ON INDUSTRIAL INFORMATICS
(2022)
Article
Computer Science, Information Systems
Kheireddine Choutri, Mohand Lagha, Souham Meshoul, Mohamed Batouche, Yasmine Kacel, Nihad Mebarkia
Summary: Human-drone interaction has gained attention in recent years. This study aims to develop a multilingual speech recognition system that allows users to control drone movements using voice commands. The results show high accuracy in English, Arabic, and Amazigh recognition, and successful implementation on a quadrotor UAV.
Article
Chemistry, Physical
Jizhong Zhao, Yuan Yao, Wentao Lei, Li Zhao, Andeng Liu, Meidan Ye, Jianyang Wu, Shihui Guo, Wenxi Guo
Summary: Researchers have developed an anti-interference self-powered acoustic fabric (ASAF) that can be used as a precise and wearable sound receiver in complex acoustic environments. The ASAF, which uses polyvinylidene fluoride (PVDF) as a vibration-sensitive layer, can record human speech at a wide range of frequencies (0-5000 Hz). The system can recognize 25 words related to extreme weather conditions with over 95.8% accuracy. The ASAF is expected to benefit outdoor rescuers, journalists, students, and other professionals working in complex acoustic environments.
Article
Biology
Xinhui Li, Xu Zhang, Xiang Chen, Xun Chen, Aiping Liu
Summary: This paper proposes a novel OT-ST framework for cross-user gesture recognition in the myoelectric control system. The framework transfers models across user domains using an unsupervised domain adaptation approach, and utilizes a teacher model and OT algorithm to improve the effectiveness of domain adaptation. Experimental results show that the OT-ST framework achieves high accuracy and outperforms other common machine learning and UDA methods.
COMPUTERS IN BIOLOGY AND MEDICINE
(2023)
Article
Chemistry, Analytical
Jesus D. Ceron, Diego M. Lopez, Felix Kluge, Bjoern M. Eskofier
Summary: Indoor localization and human activity recognition are crucial for context-based assistance in ambient assisted living scenarios. This study presents and evaluates a framework that combines indoor localization, mapping, and human activity recognition by leveraging the relationship between location and activity. The framework incorporates data from an inertial measurement unit and Bluetooth low energy beacons to achieve non-intrusive configuration. Pilot study results demonstrate high accuracy in recognizing seven activities of daily living and indoor localization, with no significant differences observed between age groups.
Article
Engineering, Electrical & Electronic
Luis Felipe Parra-Gallego, Juan Rafael Orozco-Arroyave
Summary: This paper focuses on finding suitable features to robustly recognize emotions and evaluate customer satisfaction from speech in real acoustic scenarios. The results suggest that the I2010PC feature set is the best approach to classify emotions in standard databases, while articulation features perform best in recordings from call-center settings.
DIGITAL SIGNAL PROCESSING
(2022)
Article
Chemistry, Multidisciplinary
Dan Ungureanu, Stefan-Adrian Toma, Ion-Dorinel Filip, Bogdan-Costel Mocanu, Iulian Aciobanitei, Bogdan Marghescu, Titus Balan, Mihai Dascalu, Ion Bica, Florin Pop
Summary: The evolution of Natural Language Processing technologies has made them a viable choice for various accessibility features and improving human-computer interactions. This article presents an architecture based on speech processing systems to enhance Romanian emergency services and reduce response times. The authors also release a large high-quality speech dataset for Romanian and achieve state-of-the-art results on call transcription and emotion classification. The system is designed to be integrated with the existing Romanian emergency system.
APPLIED SCIENCES-BASEL
(2023)
Article
Acoustics
Matthias Brandt, Simon Doclo, Timo Gerkmann, Joerg Bitzer
JOURNAL OF THE AUDIO ENGINEERING SOCIETY
(2017)
Article
Acoustics
Feifei Xiong, Stefan Goetze, Birger Kollmeier, Bernd T. Meyer
IEEE-ACM TRANSACTIONS ON AUDIO SPEECH AND LANGUAGE PROCESSING
(2018)
Article
Acoustics
Christina Imbery, Sven Franz, Steven Van De Par, Joerg Bitzer
JOURNAL OF THE AUDIO ENGINEERING SOCIETY
(2018)
Article
Acoustics
Feifei Xiong, Stefan Goetze, Birger Kollmeier, Bernd T. Meyer
IEEE-ACM TRANSACTIONS ON AUDIO SPEECH AND LANGUAGE PROCESSING
(2019)
Article
Acoustics
Benjamin Cauchi, Kai Siedenburg, Joao E. Santos, Tiago H. Falk, Simon Doclo, Stefan Goetze
IEEE-ACM TRANSACTIONS ON AUDIO SPEECH AND LANGUAGE PROCESSING
(2019)
Article
Audiology & Speech-Language Pathology
Inga Holube, Petra von Gablenz, Joerg Bitzer
Article
Health Care Sciences & Services
Longdan Hao, Stefan Goetze, Tourkiah Alessa, Mark S. Hawley
Summary: This systematic review and meta-analysis examines the effectiveness of computer-tailored health communication (CTHC) in increasing physical activity for individuals with or at risk of long-term conditions. The results indicate that CTHC has a significant small to medium effect in promoting physical activity compared to general health information or no information. However, further studies are needed to guide design decisions for maximizing the effectiveness of CTHC.
JOURNAL OF MEDICAL INTERNET RESEARCH
(2023)
Proceedings Paper
Acoustics
George Close, Thomas Hain, Stefan Goetze
Summary: Recent work in speech enhancement has explored the use of self-supervised speech representations (SSSRs) in the loss functions. However, the relationship between the language used to train the self-supervised representation and the language used to train the enhancement system has received little attention. This study investigates the effect of training language and amount of training data on the performance of the enhancement models across different languages. The results show that the training language of the self-supervised representation has a minor effect on performance, while the amount of training data for a specific language greatly affects performance.
2023 IEEE WORKSHOP ON APPLICATIONS OF SIGNAL PROCESSING TO AUDIO AND ACOUSTICS, WASPAA
(2023)
Proceedings Paper
Acoustics
William Ravenscroft, Stefan Goetze, Thomas Hain
Summary: This paper proposes a weighted multi-dilation temporal convolutional network (WD-TCN) for speech dereverberation task, which consistently outperforms the traditional TCN model in various model configurations, and using WD-TCN is a more parameter-efficient method.
2022 INTERNATIONAL WORKSHOP ON ACOUSTIC SIGNAL ENHANCEMENT (IWAENC 2022)
(2022)
Proceedings Paper
Acoustics
George Close, Samuel Hollands, Thomas Hain, Stefan Goetze
Summary: This paper proposes neural models for predicting Speech Intelligibility (SI) and trains and fine-tunes intrusive SI predictors on true HSR scores. The results show that neural predictors outperform non-neural baselines after fine-tuning on the true HSR scores.
Proceedings Paper
Acoustics
William Ravenscroft, Stefan Goetze, Thomas Hain
Summary: This paper analyzes the impact of model size and receptor field (RF) on the dereverberation performance of TCNs in speech processing. The experiments show that larger RF can significantly improve performance when training smaller TCN models. It is also found that TCNs benefit from a wider RF when dereverberating room impulse responses (RIRs) with larger RT60 values.
2022 30TH EUROPEAN SIGNAL PROCESSING CONFERENCE (EUSIPCO 2022)
(2022)
Proceedings Paper
Acoustics
George Close, Thomas Hain, Stefan Goetze
Summary: Training speech enhancement systems with psychoacoustically motivated speech perception metrics can improve performance and generalization.
2022 30TH EUROPEAN SIGNAL PROCESSING CONFERENCE (EUSIPCO 2022)
(2022)
Article
Acoustics
Jule Pohlhausen, Inga Holube, Joerg Bitzer
Summary: Recently, there has been a lot of attention on exploring the acoustic conditions of people in their everyday environments. This contribution proposes an algorithm to determine the own-voice audio segments (OVS) and a method for measuring sound pressure levels (SPL). The algorithm, based on machine learning and acoustic features, allows for reliable own voice detection.
Article
Engineering, Electrical & Electronic
William Ravenscroft, Stefan Goetze, Thomas Hain
Summary: This paper proposes a method to enhance the performance of speech separation networks using multihead attention mechanisms. Experimental results show that this method can significantly improve the quality and intelligibility of speech separation in noisy and reverberant environments.
FRONTIERS IN SIGNAL PROCESSING
(2022)
Article
Audiology & Speech-Language Pathology
Petra von Gablenz, Ulrik Kowalk, Jorg Bitzer, Markus Meis, Inga Holube
Summary: The study utilized Ecological Momentary Assessment (EMA) in adults with mild-to-moderate hearing loss to investigate the effects of hearing aid fitting. It found significant differences in hearing-related dimensions between first-time and experienced hearing aid users, while EMA data collected in the unaided condition did not accurately predict long-term hearing aid use.