Acoustic feature generating method and device, speech model training method and device and speech recognition method and device
An acoustic feature, a technology of a voice frame, applied in the field of data processing, can solve problems such as difficult to meet voice recognition and inaccurate recognition results, and achieve the effect of increasing the proportion, integrating the accuracy of the acoustic information vector, and improving the accuracy
Pending Publication Date: 2021-11-02
BEIJING YOUZHUJU NETWORK TECH CO LTD
View PDF10 Cites 1 Cited by
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
However, the recognition results obtained by speech model recognition are not accurate enough to meet the needs of speech recognition.
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
specific Embodiment approach
[0088] Further, the embodiment of the present application provides an integrated method that uses the integrated acoustic information vector corresponding to the previous speech frame and the acoustic information vector of the current speech frame to output and issue if the weight of the accumulated information amount corresponding to the current speech frame is greater than or equal to the threshold value. The specific implementation of the acoustic information vector specifically includes the following two steps:
[0089] A1: If the weight of the accumulated information volume corresponding to the current speech frame is greater than or equal to the threshold, calculate the weight of the accumulated information volume corresponding to the previous speech frame and multiply it by the retention rate corresponding to the current speech frame to obtain the first value, and calculate 1 and the first The difference between the numerical values is used to obtain the weight of the ...
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 embodiment of the invention discloses an acoustic feature generating method and device, a speech model training method and device and a speech recognition method and device. And according to the accumulated information amount weight corresponding to the previous speech frame, the retention rate corresponding to the current speech frame and the information amount weight of the current speech frame, the accumulated information amount weight corresponding to the current speech frame can be obtained. The retention rate is the difference between 1 and the leakage rate. The accumulated information amount weight corresponding to the current speech frame and the integrated acoustic information vector corresponding to the current speech frame are adjusted by using the leakage rate, so that the influence of the speech frame with relatively small information amount weight on the integrated acoustic information vector can be reduced; and the proportion of the acoustic information vector of the speech frame with the large information weight in the integrated acoustic information vector is increased, the obtained integrated acoustic information vector is more accurate, and the accuracy of the speech model is improved.
Description
technical field [0001] The present application relates to the field of data processing, and in particular to a method and device for generating acoustic features, training a speech model, and recognizing speech. Background technique [0002] Speech recognition technology refers to the recognition of speech data, and the content corresponding to the speech data is converted into computer-readable input. For example, through the speech recognition technology, the content included in the speech data can be converted into corresponding text, which facilitates subsequent processing of the content included in the speech data. [0003] At present, the speech recognition of speech data can be realized by using the speech model. The voice model extracts the acoustic features of the voice data, and processes the acoustic features to obtain text recognition results corresponding to the voice data. However, the recognition results obtained by the speech model recognition are not accur...
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 Patent Type & Authority Applications(China)
IPC IPC(8): G10L15/02G10L15/06G10L19/04
CPCG10L15/02G10L15/063G10L19/04G10L15/16G10L15/22
Inventor 董林昊马泽君
Owner BEIJING YOUZHUJU NETWORK TECH CO LTD