DNN (Deep Neural Network)-HMM (Hidden Markov Model)-based civil aviation radiotelephony communication acoustic model construction method

Abstract

The invention relates to a DNN (Deep Neural Network)-HMM (Hidden Markov Model)-based civil aviation radiotelephony communication acoustic model construction method. The method includes the following steps that: a Chinese radiotelephony communication corpus is set up; civil aviation radiotelephony communication speech signals are pre-processed; Fbank features are extracted from the civil aviation radiotelephony communication speech signals and are adopted as civil aviation radiotelephony communication speech features; linear discrimination analysis, feature space maximum likelihood regression transformation and speaker adaptive training transformation processing are performed on the civil aviation radiotelephony communication speech features; and the processed speech features are utilized to build a DNN-HMM-based radiotelephony communication acoustic model. With the method of the invention adopted, the FBANK and MFCC features of radiotelephony communication speech are extracted to traina DNN network, so that the DNN-HMM acoustic model suitable for radiotelephony communication speech recognition can be obtained; and since a dictionary and a language model are combined, so that the feature enhanced DNN-HMM model can reduce the phoneme recognition error rate of the radiotelephony communication speech to 5.62% on the basis of constructed data.

Description

technical field [0001] The invention belongs to the technical field of speech recognition, and in particular relates to a DNN-HMM-based method for constructing an acoustic model of a civil aviation land-air call. Background technique [0002] With the continuous development of the national economy, due to the fast and comfortable characteristics of the aircraft, it has become the preferred means of transportation for people to travel. Security will face a tougher test. As the main information communication method between pilots and controllers during the flight, civil aviation land-air calls are of great significance to ensuring flight safety. Only when controllers and pilots correctly understand the content of land-air calls can they effectively guarantee flight safety. [0003] Due to the special application scenarios, sentence structure and special pronunciation of land and air calls, the general speech recognition model cannot be applied to the field of land and air cal...

AI Technical Summary

Problems solved by technology

The hidden layers of the convolutional neural network (CNN) are not fully connected, and the feature dimension is reduced through the convolution calculation of the convolution kernel; the hidden layers of the long short-term memory network (LSTM) are fully connected, and timing information can be obtained. But the computational complexity is high

Images