Vehicle Audio Volume Control Using Passenger Voice Detection

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Passengers frequently need to adjust the audio volume in vehicles to facilitate conversation, leading to distraction and safety risks due to repeated volume adjustments.

Innovation Solution

A machine-learning model trained on audio samples to detect passenger voice, dynamically adjusting the sound volume of vehicle loudspeakers based on the presence or absence of passenger voice, using convolutional and recurrent neural networks to improve accuracy and efficiency.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Ease of operation

If passengers manually adjust audio volume to facilitate conversation, then communication between passengers is improved, but driver distraction and safety risks increase

Engineering Contradiction:
Improvecommunication convenienceVSAvoiddriving safety
Core Design Contradiction:
Ease of operationVSReliability

Solution Approach 1:

The system automatically detects passenger voice presence and adjusts audio volume without requiring manual intervention. The machine learning model continuously monitors audio samples from microphones, identifies when passengers are speaking, and autonomously controls the loudspeaker volume, making the system self-regulating and eliminating the need for driver interaction

Inventive Principle:
Principle #25Self-service

Solution Approach 2:

The system implements a closed-loop feedback mechanism where the machine learning model continuously analyzes audio input from microphones, detects passenger speech patterns, and dynamically adjusts volume based on detected speech presence. This real-time feedback loop ensures automatic adaptation to conversation needs without driver involvement

Inventive Principle:
Principle #23Feedback

2Ease of operation

If audio volume is reduced for passenger conversation, then communication is improved, but audio playback quality deteriorates when no conversation is occurring

Engineering Contradiction:
Improvecommunication convenienceVSAvoidaudio playback quality
Core Design Contradiction:
Ease of operationVSLoss of information

Solution Approach 1:

The system dynamically adjusts audio volume based on real-time detection of passenger speech. The machine learning model continuously monitors audio samples and changes volume levels adaptively - reducing volume only when speech is detected and maintaining normal levels when no conversation is occurring. This dynamic adjustment preserves audio quality during non-conversation periods while enabling communication when needed

Inventive Principle:
Principle #15Dynamics

3Measurement precision

If machine learning model continuously monitors audio samples, then voice detection accuracy is improved, but computational energy consumption increases

Engineering Contradiction:
Improvevoice detection accuracyVSAvoidcomputational energy consumption
Core Design Contradiction:
Measurement precisionVSUse of energy by moving object

Solution Approach 1:

The system processes audio samples at optimized intervals rather than continuously analyzing every audio moment. The machine learning model is trained to efficiently process discrete audio samples and make quick classification decisions, reducing computational load while maintaining detection accuracy. This periodic processing approach balances precision requirements with energy conservation in the vehicle's computing system

Inventive Principle:
Principle #19Periodic action

Data Source

PatentEP4700771A1Dynamical adjustment of sound volume inside a vehicle
Publication Date: 2026.02.25 VALEO TELEMATIK & AKUSTIK GMBH
  • EP4700771A1 patent drawingFigure 1
  • EP4700771A1 patent drawingFigure 2
  • EP4700771A1 patent drawingFigure 3

AI summary

The disclosure notably relates to a computer-implemented method of machine-learning for use in a dynamical adjustment of a sound volume of one or more loudspeakers inside a vehicle. The method comprises obtaining a dataset of audio samples. Each audio sample having a label representing a presence or absence of passenger voice. The method also comprises training a machine-learning model based on the dataset. The machine-learning is configured to receive an input audio sample. The machine-learning is configured to output information relative to presence or absence of passenger voice in the input audio sample.