Hardware-software Prototype of a Voice Navigation Subsystem in Noisy Environments

Counts differ because each database indexes a different set of publications. We treat OpenAlex as the canonical count; Google Scholar is not shown (no API, and crawling it violates its ToS).

The work is devoted to the development of the hardware-software prototype of a voice navigation subsystem intended for assistive navigation under acoustic interference. The relevance of the topic is explained by the need to improve the accuracy of generating voice or haptic prompts regarding the recommended route for visually impaired users based on their voice-based destination requests. The accuracy of the recommended route depends on the accuracy of voice command recognition in noisy and crowded environments. The goal of this paper is to develop a hardware-software module for voice navigation within an assistive system that enables user positioning and command recognition in the presence of acoustic interference, with the goal of enhancing the independence and mobility of individuals with visual and/or musculoskeletal impairments. The subject of this study is the processes and methods for improving the accuracy and robustness of voice command recognition in a hardware-software navigation subsystem in the presence of stationary and non-stationary noise and a multi-voice acoustic environment. The task of this work is to develop a functional model of the voice navigation hardware-software module, to develop a comprehensive sound processing algorithm that accounts for various types of acoustic interference and voice overlap; to analyze the impact of various types of acoustic interference and voice overlap on the performance of the software-hardware system. Methods such as comparative analysis, experimental models, acoustic signal recognition were applied. The results demonstrate that the proposed preprocessing and speaker-oriented filtering improve speech command recognition quality in all considered cases. WER decreased from 15.4% to 13.2% in reference conditions, from 48.6% to 30.9% under non-stationary noise, from 68.7% to 15.4% under 50 dB stationary noise, and from 73.0% to 35.3% for overlapping speech of equal loudness. The obtained results confirm the practical suitability of the proposed subsystem for assistive navigation systems operating in acoustically challenging real-world environments. Further research will focus on integrating intelligent personalization techniques into navigation feedback.