Multilingual identification method and system

The invention discloses a multi-language recognition method and system, and the method comprises the steps: obtaining to-be-recognized language information inputted by a user, and judging whether the to-be-recognized language information is voice information or text information; when the language information is judged to be voice information, adopting a pre-trained deep neural network model containing a bottleneck layer to carry out nonlinear transformation on the voice information so as to extract depth trend features, and respectively inputting language classification models corresponding to different language categories to carry out matching calculation, obtaining a language category corresponding to the language classification model outputting the maximum model probability score; and when the language information is judged to be text information, identifying the language category of the text information based on an N-Gram naive Bayes classification method, thereby obtaining the language category of the la