High-efficient voice instruction recognition method based on fuzzy theory and expert system

Abstract

The invention relates to a high-efficient voice instruction recognition method based on a fuzzy theory and an expert system, which mainly comprises the following steps of: A. establishing a flight knowledge base and an instruction library; B. performing fuzzy voice recognition; and C. performing reasoning judgment on voice by using a voice recognition expert system to obtain instructions. Proved by practices, by the application of the high-efficient voice instruction recognition method based on the fuzzy theory and the expert system disclosed in the invention, ten voice instructions which are mostly close to a training template are recognized through fuzzy recognition, and then the expert system is used for reasoning and judging the ten instructions one by one from the instruction mostly close to the training template to exclude unreasonable voice instructions and finally find out the instruction mostly suitable to the practice, and thus the voice recognition rate reaches more than 99 percent. Moreover, the system real-time property can be realized; the requirements of flight simulation training are met; and good training effect is achieved.

Description

technical field [0001] The present invention relates to the technical field of voice recognition, especially the voice recognition technology including human-computer interaction technology and artificial intelligence technology. Background technique [0002] Current Speech Recognition Technology: [0003] The current mainstream speech recognition technology is based on the basic theory of statistical pattern recognition. A complete speech recognition system can be roughly divided into three parts: [0004] Speech Feature Extraction: Its purpose is to extract the sequence of speech features that change over time from the speech waveform. [0005] ·Acoustic model and pattern matching (recognition algorithm): The acoustic model is usually generated through a learning algorithm from the acquired speech features. During recognition, the input speech features are matched and compared with the acoustic model (mode) to obtain the best recognition results. [0006] ·Language mod...

AI Technical Summary

Problems solved by technology

Because of regional differences, the pronunciation of the same sentence is very different between the dispatchers in the north and the south, which brings certain difficulties to speech recognition;

[0022] (2) Multiple expressions of passwords

For an instruction, the commander can have multiple expressions, such as asking for altitude, there are many ways to say "altitude", "your altitude", "current altitude", "current altitude position", etc., which makes the words of command terms The database doubles, and the increase of the lexicon will reduce the accuracy of speech recognition;

[0023] (3) Similarity of passwords

For example: "Left slope of the pressure point" and "Right slope of the pressure point", "Don't make the rising angle too large", and "Don't make the rising angle small", etc., the difference between them is not more than 20%, which brings great difficulties to speech recognition. considerable difficulty

Images