Human voice recognition algorithm

Abstract

The invention provides a human voice recognition algorithm. The human voice recognition algorithm includes the steps of S1, subjecting the volume of a speaker's voice to self-adaptive processing, carrying out recognition model training on the obtained volume of the speaker's voice and performing overall normalization processing to a same maximum value; S2, subjecting a mute area of the speaker's voice to self-adaptive processing, judging a volume value of a current speaker through mean filtering, and filtering out the mute area through a threshold value; S3, filtering and denoising backgroundsound and performing consistency processing on speaker voice data; S4, extracting voice characteristics of the speaker, and extracting a high-dimensional characteristic vector of the speaker's voice through a trained neural network algorithm model; S5, comparing and identifying the voice characteristics of the speaker with a voiceprint library, and comparing the high-dimensional characteristics extracted by the neural network algorithm model through a cosine distance so as to obtain the similarity of the speaker's characteristics. The human voice recognition algorithm directly identifies the voice characteristics of the speaker, and is low in noise and high in precision.

Description

technical field [0001] The invention belongs to the technical field of voice recognition, and in particular relates to a human voice recognition algorithm. Background technique [0002] Speech recognition technology is an information technology that converts human voices, bytes or phrases into corresponding words or symbols, or gives responses through the process of machine recognition and understanding. With the rapid development of information technology, speech recognition technology has been widely used in people's daily life. For example, when using a terminal device, voice recognition technology can be used to conveniently input information in the terminal device by inputting voice. [0003] Speech recognition technology is essentially a process of pattern recognition. The patterns of unknown speech are compared with reference patterns of known speech one by one, and the best matching reference pattern is output as the recognition result. There are many recognition m...

AI Technical Summary

Problems solved by technology

This recognition method is mainly aimed at non-specific people, and it can recognize the speech of most people. However, because it is a general-purpose acoustic model, when the user's pronunciation is not standard enough or has a local accent, this The general-purpose acoustic model cannot accurately perform matching calculations. Its substantial defect is that it can only recognize the speaker's speech content, and cannot directly identify the speaker's voice characteristics.

Images