Probabilistic Neural Network Application to Speech Recognition Problem

This article discusses an approach to solving the problem of speech recognition based on probabilistic neural networks. The problem is formulated as a problem of command recognition. Commands have equal lengths (in words). Each word position has its own set of candidates. The recognition algorithm for solving this problem was developed. The core of the algorithm is a probabilistic network that recognizes modified estimates of spectral densities. The algorithm allows for high precision of recognition, which is sufficient for the creation of a voice user interface.

1. Levonovich N.I. Voice Commands Recognition Method that Uses Special Spectral Density Transform. Modelirovanie i analiz dannikh = Modelling and Data Analysis, 2022. Vol. 12, no. 3, pp. 49–57. DOI: 10.17759/mda.2022120304. (In Russ., аbstr. in Engl.)
2. Dumoulin, Vincent, and Francesco Visin. "A guide to convolution arithmetic for deep learning." arXiv preprint arXiv:1603.07285 (2016).
3. L. S. Kuravskii, S. N. Baranov Komp'yuternoe modelirovanie i analiz dannykh. Konspekty lektsii i uprazhneniya: Ucheb. posobie. – M.: RUSAVIA, 2012. – 218 s.:
4. Ulloa, Roberto. Kivy–Interactive Applications and Games in Python. Packt Publishing Ltd, 2015.