Adaptive Speech Recognition for Cockpit Accent Compensation

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Communications between pilots and air traffic controllers are often impaired due to accent-induced anomalies, particularly when pilots encounter regional dialects and phonetic variations during international flights, leading to increased workload and difficulty in understanding instructions.

Innovation Solution

An adaptive speech recognition system that automatically switches between different regional speech databases based on the aircraft's geographical location, using an acoustic model, phonetic dictionary, and language model tailored to the specific Flight Information Region (FIR), ensuring accurate speech recognition across varying dialects.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If a single standardized speech recognition system is used, then device complexity is reduced, but speech recognition accuracy deteriorates due to accent and regional dialect variations

Engineering Contradiction:
Improvespeech recognition accuracyVSAvoidspeech database structure
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The speech database is segmented into multiple regional databases, each tailored to specific geographic regions and their corresponding accent characteristics. The system divides the global speech recognition task into region-specific sub-tasks, allowing each database to optimize for local phonetic patterns while maintaining overall system accuracy across diverse regions.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The system dynamically switches between different regional speech databases based on the aircraft's current geographic location. This dynamic adaptation allows the speech recognition system to automatically adjust its acoustic models and phonetic expectations to match the local accent characteristics, thereby maintaining high recognition accuracy without requiring manual intervention.

Inventive Principle:
Principle #15Dynamics

2Reliability

If accent-specific speech databases are used for each region, then speech recognition accuracy is improved, but system complexity increases due to multiple databases and switching mechanisms

Engineering Contradiction:
Improvecommunication reliabilityVSAvoidspeech recognition system
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The speech recognition system is designed with multi-functionality to handle multiple regional dialects and accents through a single unified architecture. The system can universally process speech from any region by automatically selecting the appropriate regional database, eliminating the need for separate dedicated systems for each region while maintaining high reliability across all areas.

Inventive Principle:
Principle #6Universality (Multi-functionality)

Solution Approach 2:

The system introduces a geographic location-based intermediary mechanism that mediates between the pilot's speech input and the appropriate regional speech database. This intermediary automatically determines the current region based on aircraft position data and routes the speech recognition process to the most suitable database, thereby managing system complexity through intelligent mediation rather than direct multi-database management.

Inventive Principle:
Principle #24Intermediary (Mediator)

3Adaptability or versatility

If regional adaptation is implemented, then adaptability to different dialects is improved, but processing time increases due to database switching and location detection

Engineering Contradiction:
Improveregional dialect adaptationVSAvoidspeech processing time
Core Design Contradiction:
Adaptability or versatilityVSLoss of time

Solution Approach 1:

The system performs preliminary actions by pre-loading and maintaining multiple regional speech databases in ready state, with their corresponding geographic boundaries pre-defined. When the aircraft enters a new region, the system can quickly switch to the appropriate database without requiring time-consuming analysis or conversion, as the adaptation data is prepared in advance based on predetermined geographic zones.

Inventive Principle:
Principle #10Preliminary action

Data Source

PatentEP2858067B1System and method for correcting accent induced speech in an aircraft cockpit utilizing a dynamic speech database
Publication Date: 2018.01.31 HONEYWELL INTERNATIONAL INC
  • EP2858067B1 patent drawingFigure 1~2
  • EP2858067B1 patent drawingFigure 3

AI summary

A system and method for recognizing speech on board an aircraft that compensates for different regional dialects over an area comprised of at least first and second distinct geographical regions, comprises analyzing speech in the first distinct geographical region using speech data characteristics representative of speech in the first distinct geographical region, detecting a change in position from the first distinct geographical region to the second geographical region, and analyzing speech in the second distinct geographical region using speech data characteristics representative of speech in the second distinct geographical region upon detecting that the aircraft has transitioned from the first distinct geographical region to the second distinct geographical region.