A voice wake-up implementation method based on a small training set to optimize the decoding network

Abstract

The invention discloses a voice wake-up implementation method based on a small training set to optimize a decoding network, comprising the following steps: S1 extracting speech intrinsic features to obtain a feature vector with obvious distinction between wake-up words and non-wake-up words; S2 combining the feature vectors to obtain The feature phoneme alignment file selects a time window according to the distribution of the wake word phonemes, classifies the mapping between features and phonemes, and obtains labeled acoustic data; S3 combines the labeled acoustic data to calculate the frame-by-frame posterior probability model S4 combines to obtain acoustic data The probability model obtains the posterior probability confidence level of the phoneme level and the re-confirmation network of the wake-up word calculation network S5; the present invention can be easily implemented on general-purpose processors such as arm and dsp through simple model training strategies and steps such as optimizing the decoding network Voice wake-up and other functions.

Description

technical field [0001] The invention relates to a voice wake-up implementation method based on a small training set to optimize the decoding network. The lightweight model refers to a small training data set, and the trained model occupies a small space, and can also be used on mobile terminals with few hardware resources. The post-decoding network is optimized to reduce the offline false wake-up rate without increasing the complexity of the algorithm. Background technique [0002] Voice is the most convenient and quick means for human beings to communicate and communicate with each other. It is the dream goal of human beings to allow machines to understand voices and perform related operations in accordance with human instructions. As a result, speech recognition technology came into being. Speech recognition technology is currently an important means of human-computer interaction, and voice wake-up is an important entrance for human-computer interaction. Smart voice devi...

AI Technical Summary

Problems solved by technology

[0008] The technical problem to be solved by the present invention is to overcome the existing overall idea of ​​voice wake-up or the method of voice recognition using a large-scale vocabulary, which needs to use a large amount of training corpus to achieve good wake-up and low false wake-up effect defects, providing A voice wake-up implementation method based on a small training set and an optimized decoder that simplifies the algorithm implementation and improves the operation efficiency under the same wake-up rate, thereby realizing the purpose of voice wake-up and easy transplantation

Images