Text standardization processing method after voice recognition

Abstract

The embodiment of the invention provides a text standardization processing method after voice recognition. The method comprises the steps of setting a text conversion matching rule set in a badcase module according to an ITN error which is collected in the badcase module and fed back by a client service; inputting the pure language text to be standardized after voice recognition into a badcase module, marking the text inversely, and outputting the text; replacing at least one character marked by an inverse text in the text output by the badcase module with a corresponding number of special symbols, wherein the special symbols are selected from symbols which cannot be converted by a neural network model; inputting the text into a binary classification neural network model, outputting a 0 / 1label sequence, and determining the confidence coefficient that the model can convert the text; determining processing texts of different rules through confidence judgment; and replacing the special symbol in the processed text by utilizing the cached word, and determining a text standardization result of the pure language text. The embodiment of the invention improves the text standardization processing speed and precision, and is suitable for large-scale data processing.

Description

technical field [0001] The invention relates to the field of text processing, in particular to a text standardization processing method after speech recognition. Background technique [0002] In order to prevent some words that should not be converted into other forms of words in speech recognition, for example, the date of 2018-08-08 is mistakenly converted into the text of 2018-08-08, usually Inverse normalization will be performed. Inverse text normalization converts content requiring inverse text normalization into a labeling problem using a simple set of rules and some hand-written grammars. For the labeling problem, a compact bidirectional LSTM (Long short term memory) is applied. Assign a label to each phonetic input mark, and get the corresponding written form segment and the start and end positions of subsequent processing; apply certain editing to generate a written form string; use the subsequent processing grammar to process the marked area. [0003] In the pr...

Images