Method and device for judging confidence of speech recognition

Abstract

The invention discloses method and device for judging the confidence of speech recognition. The method comprises the following steps of: receiving an input speech; confirming a recognition result of the input speech corresponding to a recognition grammar; confirming a competition result of the input speech corresponding to an intrinsic phoneme-decoding space; and utilizing the recognition result and the competition result to calculate the confidence of the recognition result, wherein the construction process of the intrinsic phoneme-decoding space comprises the following steps of: clustering various phonemes contained in a preliminarily-trained acoustic module to acquire an intrinsic phoneme set; retraining an acoustic module and a speech module representing various intrinsic phonemes; and constructing the intrinsic phoneme-decoding space based on the intrinsic phoneme set and the acoustic module and the speech module acquired by retraining. By clustering the phonemes to construct the intrinsic phoneme-decoding space and decoding based on the space to acquire the competition result, the invention can rapidly and effectively acquire the competition result to be commonly suitable, thereby estimating the confidence more accurately and efficiently.

Description

technical field [0001] The invention relates to the technical field of speech recognition, in particular to a method and device for judging the confidence level of speech recognition. Background technique [0002] The task of the current automatic speech recognition system is to find out the most similar recognition result (the most similar text content) of the input speech (the input speech to be recognized) within the scope of the recognition grammar (language model). 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. For this reason, the automatic speech recognition system can not only give the recognition result, but also give the credibility that the recognition result is the real content of the input voice. Judgment rejection, this process is called confidence judgment. [0003] In terms of confidence judgment, pr...

AI Technical Summary

Problems solved by technology

[0010] The above methods all have their own shortcomings. Both method (1) and method (2) define an auxiliary decoding space and decode on it to obtain an effective competition path, but the auxiliary decoding space of method (1) is defined as a given recognition grammar In the decoding space, the effectiveness of competing paths is very dependent on the recognition grammar itself, and its universality is poor; method (2) uses a full-phoneme decoding network for decoding, and takes into account timing information, language models, etc. when decoding to obtain effective competing paths important knowledge, so the effective competition path can be obtained more accurately, but the algorithm complexity is large, and it is difficult to apply to the real-time speech recognition system in many occasions; method (3) uses the waste model directly as the effective competition path, and the research focus is on waste The acoustic space modeling of the model cannot dig deep into knowledge such as time series information and language models, and the performance obtained is relatively limited

Images