Method for improving rejection capability of speech recognition system

Abstract

The invention relates to a method for improving rejection capability of a speech recognition system. The method comprises the following steps of collecting various types of noise data; classifying according to the noise types; for different types of noise, respectively training GMMs (Gauss mixed model); assembling various types of GMMs into an integral absorption model; training a statistic language model by various types of relatively random texts, and then establishing a recognition network by WFST (weighted finite state transducer) technique, which is called as an absorption network; connecting the absorption network, the absorption model and an original decoding network in parallel to form a new decoding network; enabling the input original audio frequency to pass endpoint detection and a feature extraction module, so as to generate feature vectors; and competing the feature vectors in the three parts of the decoding network according to an Viterbi algorithm, so as to generate a final recognition result, and effectively reject the noise and an out-of-vocabulary condition. The method has the advantage that on the premise of balancing the recognition efficiency, the effect of rejecting the out-of-vocabulary condition and the invalid input is well realized.

Description

technical field [0001] The invention relates to a method for improving recognition rejection ability in a speech recognition system, which is used in the technical field of command word recognition in the speech recognition system. Background technique [0002] The command word recognition system is a very important category in the current speech recognition system, which is widely used in navigation products of home appliances, vehicles, smart phones and call centers. The task of the command word recognition system is to find the most similar recognition result of the input speech within the scope of recognition grammar. Compared with the limited recognition grammar, the input speech is unlimited. When the actual content of the input speech is not within the recognition grammar range, the input is called an out-of-set word. In addition to out-of-collection words, there will be other invalid inputs such as background speech and noise. After these invalid inputs are sent to ...

AI Technical Summary

Problems solved by technology

By defining an auxiliary space and performing decoding on it to obtain an effective competition path, either the effectiveness of the competition path is very dependent on the recognition grammar itself, and the universality is poor; Taking into account important knowledge such as time series information and language models when calculating the path, the effective competitive path can be obtained more accurately, but the algorithm complexity is high, and it is difficult to successfully apply it in a speech recognition system that requires a high real-time rate.

Images