Gapochkin Artem Vladimirovich
North-Caucasus Federal University, Stavropol
Abstract. The article deals with the bases of neural networks and the general principles of constructing neuronic technology for speech recognition. The aspects of neural network optimal structure choice, the implementation of network training phase according to certain algorithm are touched upon. The main advantages of neuronic method use for speech recognition are enumerated.
Key words and phrases: распознавание речи, нейронная сеть, нейросетевые технологии, обучение сети, нейрон, speech recognition, neural network, neuronic technology, network training, neuron
Open the whole article in PDF format. Free PDF-files viewer can be downloaded .
References:
Barskii A. B. Neironnye seti: raspoznavanie, upravlenie, prinyatie reshenii. M.: Finansy i statistika, 2004. 176 s.
Sistemy iskusstvennogo intellekta: sb. stud. nauch. st. / otv. red. G. S. Kiryakova. Krasnoyarsk: IPTs KGTU, 2002. 112 s.
Uossermen F. Neirokomp'yuternaya tekhnika: teoriya i praktika / per. s angl. M.: Mir, 1992. 105 s.
Khaikin S. Neironnye seti: polnyi kurs. M.: Vil'yams, 2005. 1104 s.
Chesebiev I. A. Komp'yuternoe raspoznavanie i porozhdenie rechi. M.: OOO "Sport i Kul'tura - 2000", 2008. 128 s.
Shmeleva A. Pravda o raspoznavanii rechi // Komp'yuterPress. 1998. № 1. S. 306-316.
Haffner P., Waibel A. Multi-State Time Delay Neural Networks for Continuous Speech Recognition // Advances in Neural Information Processing Systems. San Mateo: Morgan Kaufmann Publ., 1992. Vol. 4. P. 579-588.
Русский
