Anti-false wake-up training method and device, equipment and storage medium
A training method and false wake-up technology, applied in program control devices, program control design, voice analysis, etc., can solve problems that are difficult to meet real scene requirements, affect user experience, audio wake-up, etc., and reduce the probability of false wake-up , fast speed, and the effect of reducing the amount of training data
Pending Publication Date: 2022-06-03
SOUNDAI TECH CO LTD
View PDF0 Cites 0 Cited by
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
[0003] However, if the preset threshold is easy to be exceeded, it is easy to be awakened by irrelevant audio; if the preset threshold is not easy to be exceeded, if the quality of the wake-up word is not good enough, it will not be able to wake up, which greatly affects the user experience; therefore , it is difficult to meet the needs of real scenarios if only the preset threshold is adjusted
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 Construction
[0041] The following describes in detail the embodiments of the present invention, examples of which are illustrated in the accompanying drawings, wherein the same or similar reference numerals refer to the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary, and are intended to explain the present invention and should not be construed as limiting the present invention.
[0042] In order to facilitate the understanding of the embodiments by those skilled in the art, some terms are explained:
[0043] (1) asr: Automatic Speech Recognition, automatic speech recognition technology, is a technology that converts human speech into text.
[0044] (2) MCU: Microcontroller Unit, micro control unit.
[0045] (3) Loss function: It is a function that maps the value of a random event or its related random variables to non-negative real numbers to represent the "risk...
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 an anti-false wake-up training method and device, equipment and a storage medium, and the method comprises the steps: testing an initial model, and obtaining a false wake-up audio set; labeling each of the false wake-up audios to obtain a labeling word; based on each false wake-up audio and the annotation word matched with the false wake-up audio, obtaining a target training set; and training the initial model based on the target training set to obtain a target model. According to the technical scheme, the problem that the training speed is low due to the fact that too many kinds of noise are added in order to resist false awakening when the initial model is trained is solved; and the obtained target model has the characteristic of relatively high noise immunity, the speed of training the target model is high, and the required training data volume is also greatly reduced.
Description
technical field [0001] The invention belongs to the technical field of device wake-up, and in particular relates to a training method, device, device and storage medium for anti-false wake-up. Background technique [0002] At present, in order to reduce power consumption, when the device to be awakened is in a low-power standby state, only the wake-up system works normally; when a specific sound is received, the standby system is turned on and running; the model training method for voice wake-up is: wake-up The audio of the word is a positive example and the audio of other utterances is a negative example to train the model. After the model is trained, a preset threshold is defined, the model scores the incoming sound, and if the score reaches the preset threshold, it will wake up. [0003] However, if the preset threshold is easily exceeded, it is easy to be awakened by irrelevant audio; if the preset threshold is not easily exceeded, if the quality of the wake-up word is ...
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): G06F9/4401G06N20/00G10L15/06G10L15/26
CPCG06F9/4418G06N20/00G10L15/063G10L15/26
Inventor 李良斌李智勇陈孝良
Owner SOUNDAI TECH CO LTD