Speech recognition method and system, medium, computer equipment, terminal and application

Abstract

The invention belongs to the technical field of voice information processing, and discloses a voice recognition method and system, a medium, computer equipment, a terminal and an application thereof, and the method comprises the steps: carrying out the language model modeling based on the text data of a medical field, and recognizing the voice of the medical field; performing post-processing on the recognized text to obtain a reading form, wherein the reading form comprises two post-processing tasks of punctuation prediction and text inverse standardization; adding punctuation marks representing sentence boundary information to the punctuation prediction recognition text, and transliterating the symbol pronunciation text in the reading form in the recognition text into the symbol form in the reading form through text inverse standardization. According to the invention, analysis and design are carried out according to requirements of a voice recognition system in the medical field, and a web-based website system is developed. The two functional forms of universal input and template input meet the actual input form requirements of voice recognition. The voice recognition and post-processing technology is used as system support, and a software system which is excellent in performance and user-friendly is completed.

Description

technical field [0001] The invention belongs to the technical field of speech information processing, and in particular relates to a speech recognition method, system, medium, computer equipment, terminal and application. Background technique [0002] At present: Speech recognition technology has been developed for more than 60 years from the isolated digital recognition system for specific speakers, and the technical solution has gradually developed from the initial pattern matching method to today's statistical-based model. From the mid-1990s to the beginning of this century, the development of a hybrid speech recognition framework based on Gaussian Mixture Model (GMM) and Hidden Markov Model (HMM) has enabled large-scale continuous speech recognition systems. development possible. Since the beginning of this century, with the vigorous development of deep learning, speech recognition technology based on deep learning has become the mainstream. Before 2015, speech recogni...

AI Technical Summary

Problems solved by technology

However, the construction of the rule method is often cumbersome and prone to conflicts. Therefore, the data-driven sequence-to-sequence method can combine semantic information for standardized transcription, which is an important direction for current research to solve the problem of text standardization.

[0006] Through the above analysis, the existing problems and defects of the existing technology are as follows: In text standardization, the hybrid model based on deep learning and rules has been applied very well.

However, the construction of the rule method is often cumbersome and prone to conflicts. Therefore, the data-driven sequence-to-sequence method can combine semantic information for standardized transcription, which is an important direction for current research to solve text standardization problems.

[0007] The difficulty in solving the above problems and defects is: the lack of audio data related to the medical field for the customized development of the recognition system, it is very difficult to achieve accurate speech recognition in the professional field; the results of speech recognition in the medical field lack punctuation information, pauses are random and Multi-subject crossing is frequent, adding punctuation marks that represent sentence boundary information makes it difficult to construct a punctuation prediction model; recognition in the medical field involves many complex forms such as dates, numbers, and unit symbols, and converts audio in the form of reading aloud into text in the form of reading to construct text Denormalizing the model is more difficult

Images