B.S. Goryachkin1, A.A. Andrianov2, D.V. Mozhaev3

1–3 Bauman Moscow State Technical University (Moscow, Russia)

1 bsgor@mail.ru, 2 alks.andrnv@gmail.com, 3 mdv413@mail.ru

Problem setting. Speech recognition systems are rapidly being introduced into all areas of human activity. For effective use of the speech recognition system, it is necessary to build a high-quality and productive model of human-application interaction that will facilitate the work and involve the user in the process. This paper examines the analysis of ways to improve the quality of human interaction and the cloud speech recognition system, examines the performance of the output system and its relationship with human perception of the interface, analyzes the options for interaction between the speech recognition system and the client and determines the optimal one.

Target. To analyze and compare various data transmission methods intended for processing by a cloud-based speech recognition system, to determine the most effective method based on the response rate of the user interface.

Results. It is shown how the size of the transmitted interval affects the speed of displaying recognition results. The cases of transmission of audio data broken down by words and sentences are analyzed. A comparison of technological cycles of information flows is carried out and a Gantt chart is given for each case. The regularities of the display speed depending on the length of the interval are revealed.

Practical significance. Based on the results of the analysis, it becomes possible to develop the most responsive and user-friendly system. The results of the calculations indicate a multiple increase in the response rate, which has a positive effect on the user experience.

Goryachkin B.S., Andrianov A.A., Mozhaev D.V. Analysis of non-functional requirements of the speech recognition system. Neurocomputers. 2023. V. 25. № 6. Р. 47-55. DOI: https://doi.org/10.18127/j19998554-202306-05 (In Russian)

1. Limanova N.I., Seleznev I.A. Efficiency analysis of client-server architecture. Bulletin of Science and practice. 2022. V. 8. № 7. P. 392–396. DOI 10.33619/2414-2948/80/37. (In Russian)
2. Andriyanov N.A., Kamalova Yu.B. Efficiency and performance study of computer vision algorithms in pollen grain recognition problems. Information-measuring and Control Systems. 2022. V. 20. № 5. P. 48−54. DOI 10. 18127/j20700814-202205-08. (in Russian)
3. Karpov A.A., Kipyatkova I.S. Methodology of evaluation of the work of automatic speech recognition systems. Izvestia of higher educational institutions. Instrumentation. 2012. V. 55. №. 11. P. 38–43. (In Russian)
4. Yakubov M.S., Umurzakova D.M. Methods and algorithms of speech recognition with defective pronunciation. Prospects for the development of information technologies. 2016. №. 29. P. 44–48. (In Russian)
5. Bilenko M.V., Balakshin P.V. Comparative analysis of speech recognition systems with open source. International Scientific Research Journal. 2017. № 4-4(58). P. 13–18. DOI 10.23670/IRJ.2017.58.141. (In Russian)
6. Babarinov S.L., Budnikova M.A. On speech recognition. Scientific Bulletin of Belgorod State University. Series: Economics. Computer science. 2014. № 21(192). P. 182–185. (In Russian)
7. Polyanichko K.S. Review and testing of speech recognition systems. Ogarev-Online. 2021. №. 12(165). (In Russian)