Unlock instant, AI-driven research and patent intelligence for your innovation.
Voice activity detection method and device
What is Al technical title?
Al technical title is built by PatSnap Al team. It summarizes the technical point description of the patent document.
An endpoint detection and voice technology, applied in voice analysis, speech recognition, instruments, etc., can solve the problem of low accuracy of endpoint detection and achieve the effect of ensuring accuracy
Inactive Publication Date: 2020-01-14
HUAWEI TECH CO LTD
View PDF9 Cites 14 Cited by
Summary
Abstract
Description
Claims
Application Information
AI Technical Summary
This helps you quickly interpret patents by identifying the three key elements:
Problems solved by technology
Method used
Benefits of technology
Problems solved by technology
[0006] The embodiment of the present application provides a method and device for end-of-speech detection to solve the problem of low accuracy of end-point detection in the prior art
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
Click on the blue label to locate the original text in one second.
Reading with bidirectional positioning of images and text.
Smart Image
Examples
Experimental program
Comparison scheme
Effect test
example 1
[0064] Example 1. There are overlapping voices in the voice signal. For example, the acquired text information corresponding to the voice signal is as follows.
[0065] User A: I don't know that's true;
[0066] User B: True.
[0067] Analysis: The words spoken by user A and user B include the same voice "really", and there are overlapping voices. Although the overlapping voices belong to non-linguistic information, they do not belong to the dragging information, hesitation information and delay information in non-linguistic information Any one of them, therefore, the detection duration when the speech end endpoint is detected will be processed according to the second duration in step 203.
example 2
[0068] Example 2. There are deep utterances in the voice signal. For example, the acquired text information corresponding to the voice signal is as follows.
[0069] User A: Let's wait;
[0070] User B: OK.
[0071] Analysis: The acquired voice signals show that the speaking states of both user A and user B are low-pitched, which refers to the state information of the speaker, which belongs to non-linguistic information, but does not belong to any of the drag information, hesitation information and delay information. One, therefore, the detection duration when the end point of speech is detected will be processed according to the second duration in step 203 .
example 3
[0072] Example 3. There is a speech pause in the speech signal. For example, the acquired text information corresponding to the speech signal is as follows.
[0073] User A: He drives (pause 200 milliseconds) up the hill; (turn-turn 1.3 seconds)
[0074] User B: Really? (Pause 150ms) How far?
[0075] Analysis: The acquired voice signal shows that user A pauses when he says "drive" and "up the mountain", and user B says "really" and "how far", which belongs to the delay information in non-linguistic feature information, so , the detection duration when the speech end endpoint is detected will be processed according to the first duration in step 203.
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 provides a voice activity detection method and device. The method comprises the following steps of acquiring a speech signal input by a user, and converting the speechsignal into text information; determining a context type corresponding to the text information and / or non-linguistic feature information in the speech signal; determining detection duration accordingto the context type and / or the non-linguistic feature information; recognizing a corresponding pronunciation interval of each word in the text information in the speech signal, and using the end timepoint of a corresponding pronunciation interval of a first word in the speech signal as a first end point when the pronunciation interval of a second word is determined to be not included in the detection duration after the pronunciation interval of the first word in the text information; and using the first end point as a speech end point of a sentence where the first word is located in the speech signal when the semantic structure of the sentence where the first word is located is determined to be complete.
Description
technical field [0001] The present application relates to the technical field of speech detection, in particular to a method and device for detecting a speech end endpoint. Background technique [0002] With the advancement of science and technology, people's work and life are applied to computers and networks almost every day. In order to serve work and life more conveniently and efficiently, speech recognition is widely used in various fields. For example, human-computer interactive Speech recognition, when people communicate with each other, record their conversations by voice recognition, or record their thoughts by voice anytime, anywhere, etc. This recognition method has gradually become Trends in the development of voice applications. The process of speech recognition mainly includes four steps, namely: speech signal collection, speech signal feature parameter extraction, acoustic model and pattern matching, language model and language processing. Wherein, when the ...
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
Application Date:The date an application was filed.
Publication Date:The date a patent or application was officially published.
First Publication Date:The earliest publication date of a patent with the same application number.
Issue Date:Publication date of the patent grant document.
PCT Entry Date:The Entry date of PCT National Phase.
Estimated Expiry Date:The statutory expiry date of a patent right according to the Patent Law, and it is the longest term of protection that the patent right can achieve without the termination of the patent right due to other reasons(Term extension factor has been taken into account ).
Invalid Date:Actual expiry date is based on effective date or publication date of legal transaction data of invalid patent.
Login to View More 
Login to View More