Vehicle Speech Domain Classification for Intent-Aware Recognition

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing speech recognition systems for vehicles struggle to accurately distinguish between vehicle-related and external domains, leading to inconsistent processing results due to differences in functionality among various providers, which can hinder the provision of tailored services based on user intent.

Innovation Solution

A method and system for classifying user speech into vehicle or external domains using a combination of previously stored keyword-related information and a trained domain classification model, with confidence value adjustments through re-training, to ensure accurate domain classification and appropriate service provision.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Adaptability or versatility

If a plurality of speech recognition systems are used, then service coverage and functionality are improved, but domain classification accuracy deteriorates due to inconsistent processing results

Engineering Contradiction:

Improveservice coverageVSAvoiddomain classification accuracy

Core Design Contradiction:

Adaptability or versatilityVSMeasurement precision

Solution Approach 1:

The patent segments the speech recognition service into two distinct parts: domain classification (determining whether speech is vehicle-related or external) and speech processing. By separating these functions and applying different recognition systems to different domains, the patent resolves the contradiction between service coverage and classification accuracy.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The patent introduces a domain classification module as an intermediary between the speech recognition system and the processing system. This intermediary determines whether to use the vehicle-specific recognition system or external recognition systems, thereby improving classification accuracy while maintaining service coverage through multiple systems.

Inventive Principle:

Principle #24Intermediary (Mediator)

2Adaptability or versatility

If domain classification is performed using multiple systems, then service coverage is improved, but processing consistency deteriorates

Engineering Contradiction:

Improveservice coverageVSAvoidprocessing consistency

Core Design Contradiction:

Adaptability or versatilityVSStability of the object's composition

Solution Approach 1:

The patent applies local quality by using different processing approaches for different domains: vehicle-related speech is processed with the vehicle-specific recognition system for consistency, while external speech is processed with external recognition systems for coverage. This domain-specific processing resolves the contradiction between coverage and consistency.

Inventive Principle:

Principle #3Local quality

3Speed

If keyword-based classification is used, then processing speed is improved, but classification accuracy deteriorates

Engineering Contradiction:

Improveprocessing speedVSAvoidclassification accuracy

Core Design Contradiction:

SpeedVSMeasurement precision

Solution Approach 1:

The patent applies preliminary action by first performing fast keyword-based classification to quickly identify obvious vehicle-related speech, then applying more accurate but slower trained model classification only when needed. This two-stage approach resolves the contradiction between speed and accuracy by optimizing the processing path for different cases.

Inventive Principle:

Principle #10Preliminary action

Data Source

PatentUS12494193B2Method for determining classification of a vehicle domain or an external domain based on user speech and a speech recognition system for a vehicle

Publication Date: 2025.12.09 HYUNDAI MOTOR CO LTD

AI summary

A method for determining a vehicle domain includes: converting a user's speech into text; and classifying the user's speech into a vehicle domain or an external domain based on the text, wherein the classifying of the user's speech into the vehicle domain or the external domain includes classifying a domain of the user's speech based on previously stored keyword-related information and then classifying the domain of the user's speech based on previously stored keyword-related information and then classifying the domain of the user's speech based on a trained domain classification model.