Hand Detection for Speech Recognition Activation

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Always-on automatic speech recognition (ASR) systems in mobile devices consume high power, reducing battery life, and require either a spoken activation keyword or a dedicated button, which can be inconvenient and unsafe, especially when operating a vehicle.

Innovation Solution

Activating speech recognition by detecting a user's hand positioned over a device, eliminating the need for a spoken keyword or button press, and deactivating the system when the hand is removed, thereby reducing power consumption and improving safety.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Ease of operation

If an always-on ASR system is used to continually scan audio input, then speech recognition capability is improved, but power consumption increases

Engineering Contradiction:
Improvespeech recognition capabilityVSAvoidpower consumption
Core Design Contradiction:
Ease of operationVSUse of energy by moving object

Solution Approach 1:

The system uses periodic action by implementing an always-listening mode that continuously monitors for wake words at low power, then periodically activates full ASR processing only when triggered. This allows the device to maintain speech recognition capability while significantly reducing overall power consumption compared to continuous full-power ASR operation.

Inventive Principle:
Principle #19Periodic action

2Use of energy by moving object

If a spoken activation keyword is required to enable ASR processing, then power consumption is reduced, but user convenience deteriorates due to additional time and pronunciation requirements

Engineering Contradiction:
Improvepower consumptionVSAvoiduser convenience
Core Design Contradiction:
Use of energy by moving objectVSEase of operation

Solution Approach 1:

The system extracts and removes the spoken keyword requirement from the activation process. Instead of requiring users to speak activation keywords, the device uses always-listening mode with wake word detection that automatically triggers ASR processing, eliminating the need for additional spoken commands and improving user convenience while maintaining power efficiency.

Inventive Principle:
Principle #2Taking out (Extraction)

3Ease of operation

If a dedicated button is provided to initiate speech recognition, then activation control is improved, but user safety deteriorates due to attention diversion when operating a vehicle

Engineering Contradiction:
Improveactivation controlVSAvoiduser safety
Core Design Contradiction:
Ease of operationVSObject-affected harmful factors

Solution Approach 1:

The system implements self-service by automatically detecting wake words and triggering ASR processing without requiring manual button presses. The always-listening mode enables the device to autonomously activate speech recognition when appropriate, eliminating the need for users to divert their attention to physical buttons while maintaining safe operation, particularly in driving scenarios.

Inventive Principle:
Principle #25Self-service

4Reliability

If hand detection is used to activate speech recognition, then false activations are reduced, but device complexity increases

Engineering Contradiction:
Improvefalse activation reductionVSAvoiddevice complexity
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The system uses hand detection as an intermediary mechanism to validate speech activation requests. By requiring both hand presence detection and wake word recognition before activating full ASR processing, the system reduces false activations from background noise or accidental wake word detection. This layered approach maintains reliability while managing complexity through modular sensor integration.

Inventive Principle:
Principle #24Intermediary (Mediator)

Data Source

PatentEP4004908B1Activating speech recognition
Publication Date: 2024.10.09 QUALCOMM INC
  • EP4004908B1 patent drawingFigure 1
  • EP4004908B1 patent drawingFigure 2
  • EP4004908B1 patent drawingFigure 3

AI summary

A device to process an audio signal representing input sound includes a hand detector configured to generate a first indication responsive to detection of at least a portion of a hand over at least a portion of the device. The device also includes an automatic speech recognition system configured to be activated, responsive to the first indication, to process the audio signal.