Speech recognition method based on neural network, terminal equipment and medium

Abstract

The invention is applicable to the technical field of artificial intelligence, and provides a speech recognition method based on a neural network, terminal equipment and a medium. The speech recognition method comprises the following steps: acquiring a to-be-recognized voice sequence, and dividing the speech sequence into at least two frames of speech segments; implementing acoustic characteristicextraction on the speech segments so as to obtain characteristic vectors of the speech segments; confirming a first probability vector of the speech segments on a probability calculating layer of a preset neural network model based on the characteristic vectors of the speech segments; identifying pronunciation of the speech segments as probability of a preset phoneme corresponding to each elementin the first probability vector by virtue of the value of the element; and on the basis of the first probability vectors of all speech segments in an associated time sequence classifying layer of thepreset neural network model, confirming a corresponding text sequence of the speech sequence; therefore, time cost and labor cost of speech recognition can be saved.

Description

technical field [0001] The invention belongs to the technical field of artificial intelligence, and in particular relates to a neural network-based speech recognition method, terminal equipment and a computer-readable storage medium. Background technique [0002] Speech recognition is the process of converting a sequence of speech into a sequence of text. With the rapid development of artificial intelligence technology, speech recognition models based on machine learning are widely used in various speech recognition scenarios. [0003] However, when training a traditional speech recognition model based on machine learning, for each frame of speech data in the speech sequence to be recognized, it is necessary to know the corresponding pronunciation phoneme in advance to effectively train the speech recognition model, which is It is required to frame-align the speech sequence with the text sequence before training the speech recognition model. However, the sample data used i...

AI Technical Summary

Problems solved by technology

[0004] In view of this, an embodiment of the present invention provides a neural network-based speech recognition method, a terminal device, and a computer-readable storage medium to solve the labor cost and time of existing speech recognition methods based on traditional speech recognition models. high cost problem

Images