Method for identifying speech keywords in broadcast signal
A technology for broadcasting signals and keywords, applied in speech analysis, speech recognition, instruments, etc., can solve the problems of unrecognized keywords, irrelevant vocabulary errors, recognition, etc.
Active Publication Date: 2019-09-20
SICHUAN UNIV +1
View PDF5 Cites 0 Cited by
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
[0004] The purpose of the present invention is to solve the problems in the prior art that keywords cannot be recognized and many irrelevant words are wrongly "recognized" when automatic speech recognition technology is used to detect illegal broadcasts; a method for recognizing broadcast signals is provided In the voice keyword method, the acoustic model and the text phoneme mapping table are trained by using the recorded samples of keywords. At the same time, since we do not need complete and meaningful sentences, we only need to recognize specific keywords in the voice To determine whether its nature is a normal broadcast or an illegal broadcast, which improves efficiency and simplifies the identification process
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 moreImage
Smart Image Click on the blue labels to locate them in the text.
Smart ImageViewing Examples
Examples
Experimental program
Comparison scheme
Effect test
Embodiment
[0050] Such as figure 1 As shown, a method for recognizing voice keywords in a broadcast signal includes the following steps:
[0051] Step 1: Establish an acoustic model; the acoustic model in this embodiment is a CMU sphinx model; the CMU sphinx model is a hidden Markov-Gaussian mixture model;
[0052] The joint expression formula of the Gaussian mixture model is as follows:
[0053]
[0054] Where x represents a syllable; p(x) is the probability of outputting the syllable; P(m) is the weight of the corresponding Gaussian probability density function; μ m And σ m 2 Is the parameter of the corresponding Gaussian distribution; m is the index of the sub-model, that is, the m-th sub-model; M is the total number of sub-models; N(·) is the multivariate Gaussian distribution; I is the unit matrix of the corresponding data dimension; P(x |m) is the probability of outputting the syllable for the m-th model. In addition, the complete acoustic model is designed based on the Gaussian mixtur...
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
Login to View More Abstract
The invention discloses a method for identifying speech keywords in a broadcast signal. The method comprises the following steps: 1, an acoustic model is established; 2, a character phoneme mapping table of the keywords is established; 3, a keyword sequence dictionary is established; 4, the acoustic model is trained through the keywords in the dictionary, and mapping between the speech property of the keywords and phonemes is obtained and loaded into the acoustic model; 5, mapping between the phonemes and the specified keywords is defined, and saved into the character phoneme mapping table; 6, the trained acoustic model, character phoneme mapping table and keyword sequence dictionary are loaded into a decoder; and 7, the to-be-identified broadcast signal is denoised, subjected to feature extraction, and then loaded into the decoder, and the keyword identification result is obtained. The acoustic model and the character phoneme mapping table are trained through a sample recorded by the keywords, and people do not need a complete meaningful sentence, so that the error tolerant rate of the character phoneme mapping table is increased.
Description
Technical field [0001] The present invention relates to the technical field of speech recognition, in particular to a method for recognizing speech keywords in broadcast signals. Background technique [0002] With the continuous development of social economy and radio communications, the importance of radio spectrum resources has become increasingly prominent. Many illegal broadcasts, driven by economic interests, privately purchase and erect broadcast base stations, broadcast false advertisements or engage in economic fraud, disrupt economic order, and interfere with normal communication signals, and even cause security incidents. Therefore, designing an effective method for efficient automatic screening and control of illegal broadcast signals has very important social and economic value. Traditional illegal broadcast screening mostly adopts manual listening and identification methods, which has high labor costs and is prone to errors. Although automatic speech recognition ha...
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
Login to View More IPC IPC(8): G10L15/06G10L15/08G10L15/22
CPCG10L15/063G10L15/08G10L15/22G10L2015/0631
Inventor 雒瑞森孙超武瑞娟杜淼余艳梅龚晓峰
Owner SICHUAN UNIV