A front-end processing method and system for improving far-field speech recognition

A front-end processing and speech recognition technology, applied in speech recognition, speech analysis, instruments, etc.

Active Publication Date: 2021-03-23
INST OF ACOUSTICS CHINESE ACAD OF SCI
View PDF5 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

However, reverberation varies with the acoustic characteristics of the room environment, and early reflections of different lengths have different effects on speech intelligibility. It is not a good practice to intercept the first 50ms for different reverberation times.

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • A front-end processing method and system for improving far-field speech recognition
  • A front-end processing method and system for improving far-field speech recognition
  • A front-end processing method and system for improving far-field speech recognition

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0042] Embodiments of the present application are described in detail below, examples of which are shown in the drawings, in which the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below by referring to the figures are exemplary, and are only for explaining the present application, and should not be construed as limiting the present application.

[0043] figure 1 It is a flowchart of a front-end processing method for improving far-field speech recognition provided by an embodiment of the present application. Such as figure 1 The specific implementation steps of the front-end processing method for improving far-field speech recognition shown are as follows:

[0044] Step S102, calculate the room impulse response signal, obtain the division time point of the early reverberation signal and the late reverberation signal, and intercept the direct sound signal and the ea...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

PUM

No PUM Login to View More

Abstract

The invention provides a front-end processing method and system for improving far-field voice recognition. The method comprises the steps that a space impulse response signal is calculated to obtain atime diving point of an early reverberation signal and a later reverberation signal, and a direct sound signal and the early reverberation signal are intercepted; the direct sound signal, the early reverberation signal and a clean voice signal in a voice library are subjected to convolution on the time domain, and a time domain target signal is obtained; the time domain target signal and other signals besides the time domain target signal in time domain mixed signals are respectively calculated, energy of the target signal and other signals is obtained, and ideal specific value masking is obtained through the energy of the target signal and other signals; after the time domain mixed signals are converted into frequency domain mixed signals, the amplitude of the frequency domain mixed signals and the ideal specific value masking are subjected to multiplying, and a phase position of the frequency domain mixed signals is utilized for obtaining a reconstruction signal. A target signal isseparated from mixed voice under the noise reverberation condition through ideal amplitude masking.

Description

technical field [0001] The invention relates to the field of audio signal processing, in particular to a front-end processing method and system for improving far-field speech recognition. Background technique [0002] With the continuous development of voice technology, the application of voice interaction has been very extensive, ranging from national military applications to personal applications in households. At present, there are more and more applications based on voice recognition, such as smart home, service robots, etc., but in real voice interaction scenarios, background noise and room reverberation will interfere with the transmission of voice, and these interferences not only affect voice quality and voice intelligibility The degree is damaged, and the harm to speech recognition is also great. Therefore, it is particularly important for speech recognition to separate speech from these disturbances. [0003] Based on the study of auditory masking phenomenon, Ide...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

Application Information

Patent Timeline
no application Login to View More
Patent Type & AuthorityPatents(China)
IPC IPC(8): G10L15/22G10L21/0208G10L21/0272
CPCG10L15/22G10L21/0208G10L21/0272G10L2021/02082
Inventor李军锋高飞颜永红
OwnerINST OF ACOUSTICS CHINESE ACAD OF SCI