Looking for breakthrough ideas for innovation challenges? Try Patsnap Eureka!

Historical voice frequency noise detection and elimination method

A technology of audio noise and history, applied in speech analysis, instruments, etc., can solve problems such as high complexity, dependence on the accuracy of model coefficients, and low efficiency

Inactive Publication Date: 2010-11-10
SHANGHAI CONSERVATORY OF MUSIC +1
View PDF0 Cites 18 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0003] "An Improved Wavelet Transform Threshold Denoising Method" ("Journal of Communications" Peng Yuhua) described the wavelet transform threshold denoising method, "Impulse Noise Removal Method Based on Wavelet Transform Modulus Maxima" ("Journal of Shandong University of Technology "(Natural Science Edition) Pan Jinfeng) proposed an impulse noise removal method based on wavelet transform modulus maxima, but both methods will cause changes in the original signal timbre
There is also a linear prediction method based on the AR model for impulse noise suppression. This method needs to train the signal, has high complexity and depends on the accuracy of the prediction of the model coefficients.
The Bayesian method can suppress this kind of noise better, but the algorithm is complex, the efficiency is low, and the statistical characteristics of the noise need to be estimated in advance

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
  • Historical voice frequency noise detection and elimination method
  • Historical voice frequency noise detection and elimination method
  • Historical voice frequency noise detection and elimination method

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0032] A method for detecting and eliminating historical audio noise, the method comprising the steps of:

[0033] (1) Sound modeling: Sound signals can be described by the following methods:

[0034] y(k)=x(k)+j(k)*d(k) (1)

[0035] Among them, y(k) is the polluted noisy signal, x(k) is the pure signal, j(k)*d(k) is the noise part, and j(k) is the flag bit, indicating whether there is a pulse here Noise, d(k) represents the amplitude value of impulse noise;

[0036](2) Short-time Fourier transform and spectrogram: Use a time-sliding analysis window to window and truncate the non-stationary signal to decompose the non-stationary signal into a series of approximately stable short-term signals, and then use Fourier transform to analyze each short-time stationary signal The spectrum of the signal; its definition is as follows:

[0037] STFT ( t , ω ) = ∫ - ...

Embodiment 2

[0047] According to the historical audio noise detection and elimination method described in Embodiment 1, the detection of the impulsive noise includes: (1) carrying out short-time Fourier transform to the noisy signal, (2) applying selection criteria in the frequency domain to detect the noise containing The window of the impulse noise, (3) returns the position of the signal in the time domain according to the joint time-frequency domain (n, ω).

[0048] When performing STFT on the source signal, the window type and window size have a great influence on the spectrogram. Compared with rectangular windows, Hamming window and Hanning window have better analysis performance; if the window is too large, it will cause inaccurate detection position and reduce the effective detection rate (see Section 2.4); if the window is too small , it will greatly increase the amount of computation, and at the same time lose part of the information in the window.

[0049] In order to save the e...

Embodiment 3

[0053] According to the historical audio noise detection and elimination method described in embodiment 1, the described impulse noise detection performance analysis includes using effective detection rate (Efficient Detection Percentage, abbreviated as EDP) and detection success rate (Right Detection Percentage, abbreviated as RDP), as a performance evaluation index for the impulse noise detection process. The calculation formulas of the two are as follows:

[0054] EDP ​​= (number of detected impulse noise / total number of detected signals)*100% (4)

[0055] RDP = (Number of Impulse Noise Detected / Total Impulse Noise in Signal)*100% (5)

[0056] Among them, EDP represents the accuracy of detection; RDP reflects the accuracy of detection, and obtaining the maximum RDP is the primary goal of the algorithm. The larger the EDP, the smaller the redundancy, the larger the RDP, the more accurate the detection, the better the detection performance, and the more beneficial it is for...

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 discloses a historical voice frequency noise detection and elimination method. The existing historical voice frequency noise elimination method has high complexity, depends on the veracity of predicting model coefficients, has a complex algorithm and low efficiency and needs to estimate the statistical property of the noise in advance. The method of the invention comprises: (1) sound modeling: a sound signal can be described with the following method: y(k)=x(k)+j(k)*d(k); (2) short time discrete Fourier transform and spectrogram: an analysis window sliding along time is adopted for windowing and cutting off a non-stable signal so as to decompose the non-stable signal into a series of approximatively stable short signal, and Fourier transform is adopted to analyze the frequency spectrum of each stable short signal; (3) pulse noise detection; (4) pulse noise detection performance analysis; (5) signal repair and reconstruction; and (6) experiment result and analysis. The invention is used for removing the pulse noise in the audio data.

Description

Technical field: [0001] The invention relates to a noise detection and removal method. Background technique: [0002] Audio noise reduction refers to extracting the original sound signal as pure as possible from the noisy sound signal. In audio signals, short-duration, discontinuous, and large-amplitude pulses or noise spikes are called impulse noise. Impulse noise can be generated by a variety of noise sources, such as scratches on old records, dust and particles on the surface of the record, sudden stops of the tape, and irregular changes in the surface of the record. Impulse noise is divided into short-term impulse noise and transient impulse noise (or called continuous impulse noise) according to the length of duration. The traditional signal analysis and processing tool is the Fourier transform, but the Fourier transform cannot give the law of the frequency spectrum of the signal changing with time. In recent years, the joint time-frequency analysis method has been d...

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
IPC IPC(8): G10L21/02G10L21/0232
Inventor 张潇吴粤北董笑菊朱俊敏王旌阳袁征
Owner SHANGHAI CONSERVATORY OF MUSIC
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Patsnap Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Patsnap Eureka Blog
Learn More
PatSnap group products