Hybrid Speech Recognition System for Mobile Devices

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Mobile devices face limitations in performing complex speech recognition tasks due to limited computational, memory, and battery resources, leading to delayed user experiences as these tasks are often offloaded to network servers, preventing users from viewing query results based on local data and causing significant delays.

Innovation Solution

A method and system that enables voice commands on client devices by translating audio streams into machine-readable commands using local speech recognizers, querying both local and remote databases simultaneously, and displaying results on the device, allowing for concurrent processing and display of query results from both sources.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If speech recognition tasks are offloaded to network servers, then speech recognition accuracy is improved, but response time deteriorates

Engineering Contradiction:
Improvespeech recognition accuracyVSAvoidresponse time
Core Design Contradiction:
Measurement precisionVSLoss of time

Solution Approach 1:

The speech recognition system is segmented into two parts: a lightweight recognizer embedded in the mobile device for immediate local recognition, and a comprehensive recognizer on the network server for accurate complex task recognition. This segmentation allows simple commands to be processed locally (fast response) while complex commands are handled by the server (high accuracy), resolving the contradiction between speed and accuracy.

Inventive Principle:
Principle #1Segmentation

2Adaptability or versatility

If speech recognition tasks are performed on network servers, then recognition capability is improved, but user experience deteriorates due to delay

Engineering Contradiction:
Improverecognition capabilityVSAvoiduser experience
Core Design Contradiction:
Adaptability or versatilityVSEase of operation

Solution Approach 1:

The system performs preliminary speech recognition action locally on the mobile device before potentially involving the network server. The embedded recognizer immediately processes the speech input and provides preliminary results, so users don't have to wait for server processing. This preliminary local action improves user experience by providing immediate feedback while maintaining the server's comprehensive recognition capability.

Inventive Principle:
Principle #10Preliminary action

3Loss of time

If speech recognition is performed locally on mobile devices, then response time is improved, but recognition accuracy deteriorates

Engineering Contradiction:
Improveresponse timeVSAvoidrecognition accuracy
Core Design Contradiction:
Loss of timeVSMeasurement precision

Solution Approach 1:

The system dynamically determines whether to use the embedded recognizer or the network server recognizer based on the complexity of the speech task. Simple commands are processed locally by the embedded recognizer for fast response, while complex commands are transferred to the network server for accurate recognition. This dynamic adaptation resolves the contradiction by optimizing the recognition path based on real-time requirements.

Inventive Principle:
Principle #15Dynamics

Data Source

PatentEP3477637B1Integration of embedded and network speech recognizers
Publication Date: 2021.08.11 GOOGLE LLC
  • EP3477637B1 patent drawingFigure 1
  • EP3477637B1 patent drawingFigure 2
  • EP3477637B1 patent drawingFigure 3

AI summary

A method, computer program product, and system are provided for performing a voice command on a client device. The method can include translating, using a first speech recognizer located on the client device, an audio stream of a voice command to a first machine-readable voice command and generating a first query result using the first machine-readable voice command to query a client database. In addition, the audio stream can be transmitted to a remote server device that translates the audio stream to a second machine-readable voice command using a second speech recognizer. Further, the method can include receiving a second query result from the remote server device, where the second query result is generated by the remote server device using the second machine-readable voice command and displaying the first query result and the second query result on the client device.