Speech decoding method and device and storage medium

Abstract

The invention discloses a speech decoding method and device and a storage medium, and belongs to the technical field of speech recognition. The speech decoding method includes the steps that a targettoken corresponding to a minimum decoding score is obtained from a first token list, wherein the first token list includes the first token obtained by decoding a previous audio frame, and the first token includes state pairs formed by decoding in different decoding networks and the decoding scores thereof; pruning parameters when decoding a current audio frame is determined according to the targettoken and an acoustic vector of the current audio frame; and according to the first token list, the pruning parameters and the acoustic vector, the current audio frame is decoded. According to the speech decoding method and device and the storage medium, a decoding network corresponding to a high-order language model does not need to be generated, and decoding is performed based on the decoding networks corresponding to a low-level language model and a differential language model, so that computing resources and storage resources are saved under the premise of ensuring decoding accuracy; andthe current audio frame is decoded according to decoding results of the previous audio frame, so that the decoding speed is increased.

Description

technical field [0001] The invention relates to the technical field of voice recognition, in particular to a voice decoding method, device and storage medium. Background technique [0002] Speech recognition technology, also known as ASR (Automatic Speech Recognition, automatic speech recognition), its goal is to convert the vocabulary content in human speech into computer-readable input, including keystrokes, binary codes or character sequences, etc., so as to realize human-machine interact. Speech recognition technology has a wide range of application scenarios in modern life, and can be applied to scenarios such as car navigation, smart home, voice dialing, and simultaneous interpretation. The decoder is the core of the speech recognition system, and the speech decoding process based on the decoder plays an important role in the entire speech recognition process, directly affecting the accuracy of the recognition results. [0003] At present, the speech decoding process...

AI Technical Summary

Problems solved by technology

[0004] However, the memory of the high-level language model is relatively large, and the memory of the decoding network generated based on the high-level language model is much larger than that of the high-level language model, which requires the configuration of a large number of storage resources and computing resources. In scenarios with limited computing resources, it is difficult to achieve decoding. Therefore, there is an urgent need for a speech decoding method that takes into account both decoding speed and decoding accuracy.

Images