Russian-Chinese speech translation method, Russian-Chinese speech translation device and storage medium

Abstract

The invention relates to a Russian-Chinese speech translation method, a Russian-Chinese speech translation device and a storage medium. The Russian-Chinese speech translation method comprises the following steps: acquiring to-be-translated Russian speech, and converting the to-be-translated Russian speech into a to-be-translated Mel spectrogram; and translating the Mel spectrogram to be translated into a target Mel spectrogram through a pre-trained Russian-Chinese speech translation model. and obtaining the Chinese speech corresponding to the to-be-translated Russian speech according to the target Mel spectrogram. According to the Russian-Chinese speech translation method provided by the invention, the to-be-translated Russian speech is converted into the to-be-translated Mel spectrogram, the speech features of the to-be-translated Russian speech can be accurately represented, and then the to-be-translated Russian speech is translated into the target Mel spectrogram by adopting the trained Russian-Chinese speech translation model; and when the Chinese speech corresponding to the to-be-translated Russian speech is obtained, the problem of serious accuracy loss can be reduced, and the translation quality can be improved. The Russian-Chinese speech translation rate can be increased, and the Russian information processing capability can be improved.

Description

technical field [0001] The present disclosure relates to the technical field of speech recognition, and in particular to a Russian-Chinese speech translation method, a Russian-Chinese speech translation device and a storage medium. Background technique [0002] Existing voice translation systems are mainly aimed at languages ​​such as English, Japanese, and German, which have a wide range of applications and a large number of users, but less voice translations for ethnic languages ​​and minor languages. [0003] In related technologies, there are mainly two methods for Russian-Chinese speech translation. The first is to perform the corresponding translation and speech synthesis after the translated Russian is recognized. However, in this way, after multi-step processing such as speech recognition, translation and speech synthesis, the loss rate is serious, which may easily lead to poor quality of the final Chinese speech translation obtained through translation. The other ...

AI Technical Summary

Problems solved by technology

However, with this method, after multi-step processing such as speech recognition, translation, and speech synthesis, the loss rate is serious, which may easily lead to poor quality of the final Chinese speech translation obtained through translation.

The other mainly uses manual translation, which leads to slow processing of information in the voice, and due to the lack of translators, it is easy to cause a large amount of important information not to be processed in time, resulting in the loss of its value

Images