Voice processing method, system and device and storage medium

Abstract

The invention provides a voice processing method, system and device and a storage medium, and the method comprises the steps: carrying out the limiting processing of an initial user recorded voice, and obtaining the limited user recorded voice; carrying out incomplete noise reduction processing on the limited user recorded voice, and obtaining the user recorded voice after noise reduction; disambiguating the de-noised user recorded voice to obtain a disambiguated user recorded voice; adjusting the number of factors in each frame of voice pre-framed in the disambiguated user recorded voice according to the average phoneme number of the initial user recorded voice, and obtaining each frame of adjusted voice; and according to the average volume of the voice recorded by the initial user, balancing the adjusted volume of each frame of voice, obtaining each frame of voice after balancing, and outputting each frame of voice after balancing. According to the invention, the user can record the voice more smoothly, and the auditory feeling of the listener is improved.

Description

technical field [0001] The present invention relates to the technical field of speech processing, and in particular, to a speech processing method, system, device and storage medium. Background technique [0002] When users make voice recordings and voice calls through mobile devices, due to noise in the radio environment, unclear pronunciation, and intermittent speech, the recorded voices may be indistinguishable, indistinguishable, and pause frequently. not good. At present, the industry's voice processing focuses on voice enhancement solutions such as noise reduction and echo cancellation, with the purpose of reducing the impact of ambient sound on human voices, and has made great progress. However, this processing mainly focuses on restoring a clean recording. environment, lack of optimization of the speaker's voice, ignoring the listener's hearing experience. Human pronunciation has its own pronunciation characteristics, such as speed of speech, pause habits, volume c...

AI Technical Summary

Problems solved by technology

[0002] When the user performs voice recording and voice calls through mobile devices, due to the noise in the radio environment, unclear pronunciation, intermittent speech, etc., the recorded voice may appear inconspicuous, difficult to distinguish, and frequent pauses, which will make the listener feel uncomfortable. bad

At present, the industry's speech processing focuses on speech enhancement solutions such as noise reduction and echo cancellation. The purpose is to reduce the impact of environmental sounds on human voices, and has already achieved full development. However, this processing mainly focuses on restoring a clean recording. Environment, lack of optimization of the speaker's voice, ignoring the hearing experience of the listener

Human pronunciation has its own pronunciation characteristics, such as speed of speech, habit of pauses, volume changes, etc. At present, the industry is lacking such solutions to optimize the listener's sense of hearing

Images