Подробное описание документа

Devyatkin D. D.
   Voice Command Recognition Using Deep Learning / Devyatkin D. D. // Наука, технологии и бизнес : материалы 6-ой Межвузовской конференции аспирантов, соискателей и молодых учёных, Москва, 16-18 апреля 2024 года / МГТУ им. Н. Э. Баумана (национальный исследовательский университет). - М., 2024. - С. 128-135.

In this work, the architecture of a deep neural network, which solves the problem of recognizing voice commands for controlling a quadcopter model, is being developed. Also, to teach the developed architecture, the dataset is collected, and its intelligent analysis and prepreparation for solving the task are carried out. As a result, the solution to the problem is reduced to an approach based on solving the classification problem. A special feature of this model is the recognition of the entire command. Neural network modeling is performed in the Python programming language using the frameworks pytorch, numpy, pandas. To speed up the training of models, the NVIDIA A100 graphics accelerator was used. After training the models, their qualitative analysis is performed which includes evaluation of metrics, values of the loss function and quality assessment in the context of the problem being solved.
Keywords: Deep Learning, Data analysis, Sound processing, Classification task