Occupant-Specific Audio Filtering in Vehicles

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Modern vehicles face challenges in distinguishing speech signals from multiple individuals speaking simultaneously or in rapid succession, which adversely affects speech recognition performance.

Innovation Solution

A method and system that utilize a position sensor to determine the positions and speech activity of occupants within a defined space, combined with microphones and processors to apply beamformers and time-frequency masks to separate audio signals and generate output signals corresponding to each occupant, effectively filtering sound and enhancing speech recognition.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If multiple microphones are used to receive sound from multiple occupants, then the ability to capture speech signals improves, but the difficulty of distinguishing and separating individual speech signals worsens

Engineering Contradiction:

Improvespeech signal separation accuracyVSAvoidspeech signal distinction difficulty

Core Design Contradiction:

Measurement precisionVSDifficulty of detecting and measuring

Solution Approach 1:

The patent segments the mixed audio signal into individual occupant speech signals by applying temporal-spatial filters to separate the speech of multiple occupants based on their spatial positions and temporal characteristics, resolving the difficulty of distinguishing individual speech signals

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The patent introduces temporal-spatial filters as an intermediary processing layer between the microphones and speech recognition system, which mediates the separation of mixed speech signals by exploiting spatial and temporal differences in the audio signals from different occupants

Inventive Principle:

Principle #24Intermediary (Mediator)

2Adaptability or versatility

If speech recognition system processes mixed speech signals from multiple occupants, then the functionality for handling multiple speakers improves, but the recognition accuracy deteriorates

Engineering Contradiction:

Improvemulti-speaker handling capabilityVSAvoidspeech recognition accuracy

Core Design Contradiction:

Adaptability or versatilityVSMeasurement precision

Solution Approach 1:

The patent segments the mixed speech signal into separate individual speech signals corresponding to each occupant before processing, allowing the speech recognition system to maintain high accuracy by processing separated signals rather than mixed signals

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The patent performs preliminary separation of speech signals using temporal-spatial filtering before the speech recognition process, preparing distinct speech signals for each occupant in advance to ensure accurate recognition

Inventive Principle:

Principle #10Preliminary action

Data Source

PatentUS9390713B2Systems and methods for filtering sound in a defined space

Publication Date: 2016.07.12 GM GLOBAL TECHNOLOGY OPERATIONS LLC

AI summary

Methods and systems are provided for filtering sound. A position sensor determines positions of a plurality of occupants in a defined space. Multiple microphones receive sound and generate corresponding audio signals. A processor in communication with the microphones and the position sensor receives the positions of the occupants and the audio signals. The processor determines which of the occupants are engaging in speech and applies a temporal-spatial filter to the audio signals to generate a plurality of output signals corresponding respectively to each occupant of the defined space.