Projects

QNX Sound: Redefining In-Vehicle Audio for the Software-Defined Era

ISD Team

Abstract

QNX Sound is a digital audio platform for software-defined vehicles that decouples audio and acoustics software from hardware, unifying all vehicle functionalities, including voice treatment, media playback, alerts and noise control into one system, streamlining development, reducing hardware needs, and speeding time-to-market for automakers.

Description

QNX Sound is a groundbreaking digital audio platform, purpose-built for software-defined vehicles, based on decades of experience delivering vehicle acoustics technology. As the industry moves to centralized computing and dynamic software architectures, QNX Sound decouples audio and acoustics software from vehicle hardware, consolidating it into a single, unified platform. Now, OEMs can reduce hardware dependencies and accelerate time-to-market for new audio features while allowing for remote collaboration / testing in the cloud, significantly improving development efficiency.

How?

QNX native audio framework for general purpose CPUs, which processes audio signals in real time on the QNX OS.
Full compatibility with VirtIO, which integrates audio processing in Android-based digital cockpit platforms running on QNX Hypervisor.
Intelligent tuning tools purposefully built for auto applications that allow seamless connectivity with several HW targets and the cloud.
A plugin that extends QNX Sound’s functionality with signal processing algorithms from 3rd parties like Dirac.

In July 2024, QNX Sound was presented at the AES conference for automotive audio in a demo that for the first time ever showed a Software Defined Audio implementation of Dolby ATMOS content reproduction. Dolby ATMOS content was reproduced by an Android app, transported via VirtIO standard to QNX Hypervisor and post-processed natively in QNX Sound using Dirac signal processing plugin. The resulting audio signals were then transported via A2B bus to the booster amplifier and then to the vehicle speakers. No DSPs were used in this implementation, and only 2% of the CPU was consumed.