Unlock instant, AI-driven research and patent intelligence for your innovation.

A method and device for automatic switching of speech recognition engines

A speech recognition engine and automatic switching technology, applied in speech recognition, speech analysis, instruments, etc., to improve efficiency and facilitate expansion

Active Publication Date: 2020-06-30
深圳如布科技有限公司
View PDF8 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0003] The purpose of the present invention is to provide a method for automatic switching of speech recognition engines, which can overcome the defect that the existing speech recognition system can only realize one language 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 more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • A method and device for automatic switching of speech recognition engines
  • A method and device for automatic switching of speech recognition engines

Examples

Experimental program
Comparison scheme
Effect test

example 1

[0063] Carry out syntactic analysis to speech text in step 31, obtain following result:

[0064] Subject: I, predicate: think, object: learn English.

[0065] Matching the predicates and objects obtained by syntactic analysis with the intention library, the following probabilities can be obtained:

[0066] P(learning scenario︱want to learn English)=1

[0067] P(translation scenario︱want to learn English)=0.5

[0068] P(Music scene︱I want to learn English)=0

[0069] where P(A︱B) represents the probability of event A occurring given event B.

[0070] By comparing the values ​​of the above conditional probabilities and determining that the maximum value is P(Learning Scenario︱I want to learn English), it can be determined that the application scenario corresponding to the current user is the "Learning Scenario".

[0071] In the subsequent execution process, after the grammatical components in the speech text are marked through syntactic analysis, the grammatical components a...

example 2

[0073] The default speech recognition engine preset by the method is the Chinese speech recognition engine. The default speech recognition engine is generally preset according to the product sales market. For example, if the product sales market is China, then the default speech recognition engine is the Chinese speech recognition engine, and the product sales market is the United States, then the default speech recognition engine is the English speech recognition engine .

[0074] The current speech recognition engine is an English speech recognition engine, and the application scenario is a learning scenario. When the user says "goodbye", the speech text "goodbye" is matched with the learning scene intention library associated with the learning scene through step 3, and the following probabilities can be obtained:

[0075] P(Chinese︱goodbye)=1

[0076] P(German︱goodbye)=0

[0077] P(Russian︱goodbye)=0

[0078] By comparing the values ​​of the above conditional probabilit...

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

A speech recognition engine automatic switching method and a device are provided; the method comprises the following steps: collecting user inputted a first audio signal, and processing the first audio signal; carrying out speech recognition for the processed first audio signal, and thus obtaining a corresponding speech text; carrying out meaning analysis for the speech text, and determining a user corresponding language kind tendency; switching the speech recognition engine according to the language kind tendency, thus carrying out speech recognition for the following second audio signal. The method can automatically switch different speech recognition engines according to user language kind tendency.

Description

technical field [0001] The present disclosure relates to the field of speech recognition, in particular to a method and device for automatic switching of speech recognition engines. Background technique [0002] Speech recognition is the basis for the realization of artificial intelligence. Many robots, Internet of Things, and mobile smart terminals use voice as an interactive portal. The speech recognition engine collects the audio signal of the client, then sends it to the speech recognition engine for recognition and analysis of the audio signal, and finally converts the audio signal into textual information to achieve the purpose of speech recognition. At present, most speech recognition systems use a single-engine working mode, that is, they can only recognize one language at work. For example, the Lingxi of HKUST Xunfei can only recognize Chinese. If the user speaks English to Lingxi, it will not be recognized correctly. This results in speech recognition systems that...

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 & Authority Patents(China)
IPC IPC(8): G10L15/00G10L15/18G10L15/26G10L21/0208
CPCG10L15/005G10L15/1822G10L15/26G10L21/0208G10L2021/02082
Inventor 原树旗雷宇
Owner 深圳如布科技有限公司