Cloud Server Voice Wakeup Re-Recognition Logic

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing voice wakeup methods for smart devices often result in command recognition errors when wakeup words are input multiple times, leading to mistaken identification of subsequent voice requests, which fails to satisfy user expectations and improves user experience.

Innovation Solution

A voice wakeup method that involves receiving and processing voice inputs at a cloud server, where the server performs voice recognition and semantic parsing to identify pre-stored wakeup words, determining if the input is for initial wake-up or re-wake-up, and sending appropriate instructions to the smart device to prevent service satisfaction processing for re-wakeup requests.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If the smart device uses voice wakeup words to wake up from standby state, then the device can enter voice recognition state for user interaction, but subsequent wakeup words are mistakenly recognized as voice requests causing command recognition errors

Engineering Contradiction:
Improvecommand recognition accuracyVSAvoidvoice input processing complexity
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The system performs preliminary voice recognition and semantic parsing on the first voice input to determine whether it is a wakeup word or a service request before proceeding with service satisfaction processing. This preliminary classification prevents misidentification of subsequent wakeup words as service requests.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The system uses feedback from the voice recognition and semantic parsing results to adjust its processing behavior. When wakeup words are detected after the device has entered voice recognition state, the system stops service satisfaction processing and sends a re-wakeup instruction, thereby correcting the potential misidentification error.

Inventive Principle:
Principle #23Feedback

2Productivity

If the cloud server performs service satisfaction processing for the first voice request, then user service can be provided, but if the request was actually a wakeup word, incorrect service is provided

Engineering Contradiction:
Improveservice response efficiencyVSAvoidvoice request type identification accuracy
Core Design Contradiction:
ProductivityVSMeasurement precision

Solution Approach 1:

The cloud server performs preliminary voice recognition and semantic parsing on the received voice request to identify whether it is a wakeup word or a service request before initiating service satisfaction processing. This preliminary identification ensures accurate classification and prevents incorrect service provision.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The system uses feedback from the semantic parsing results to control the service satisfaction processing flow. When wakeup words are identified, the system stops service satisfaction processing and sends a re-wakeup instruction to the smart device, thereby preventing incorrect service execution.

Inventive Principle:
Principle #23Feedback

Data Source

PatentEP3486901B1Voice wakeup method, cloud server and computer-readable medium
Publication Date: 2023.06.14 BAIDU ONLINE NETWORK TECH (BEIJIBG) CO LTD
  • EP3486901B1 patent drawingFigure 1
  • EP3486901B1 patent drawingFigure 2
  • EP3486901B1 patent drawingFigure 3

AI summary

The present disclosure provides a voice wakeup method, apparatus and system, a cloud server and a readable medium. The method comprises: receiving a first request input by the user in a voice manner after a smart device enters a voice recognition phase; judging whether the first request after semantic parsing is consistent with pre-stored wakeup words; determining that the received first request is used to re-wake up the smart device if the first request is consistent with the pre-stored wakeup words. According to the above technical solutions, after the smart device enters the voice recognition phase, if the received first request is the wakeup words, the smart device does not perform service satisfaction processing for the first request. Capability of recognizing that the first request is re-wakeup does not cause a command recognition error and thereby can satisfy the user's expectation. The voice wakeup solution according to the present disclosure can effectively process the user's request and thereby can effectively improve the user's experience in use.