Immersive audio experiences are increasingly created, distributed, and consumed using representations based on Audio Objects rather than fixed audio channels. In an object-based representation, each Audio Object carries audio content together with metadata describing its properties, behaviour, and relationship to the surrounding Audio Scene. This approach enables a listening experience that can be adapted to different rendering environments, devices, user preferences, and points of view.
However, the availability of Audio Objects and Audio Scene Descriptors alone does not guarantee that a satisfactory listening experience can be delivered to an end user. The final audio presentation depends on the rendering process, which must take into account the characteristics of the Audio Scene, the acoustic environment, the rendering platform, and user commands that may modify the presentation of the scene in real time. Different users may require different renderings of the same scene, for example because they wish to change the listener position, emphasise a dialogue source, modify the balance among objects, or adapt the reproduction to a specific listening environment.
Technical Specification: Context-based Audio Enhancement (MPAI-CAE) – Audio Object Rendering (CAE-AOR) V1.0 addresses this need by specifying a standard framework for rendering Audio Objects described by Audio Scene Descriptors into an audio presentation suitable for human listening. The standard provides AI-based components and associated Data Types enabling interoperable implementations capable of generating a rendered audio experience from object-based audio scene descriptions. The specification supports the dynamic adaptation of the rendered scene according to contextual information and User Commands.
CAE-AOR is designed to operate within the MPAI-specified AI Framework (MPAI-AIF) standard, where processing functions are implemented as AI Modules (AIMs) exchanging standardised Data. By relying on common MPAI frameworks and data representations, CAE-AOR promotes interoperability among content creators, service providers, application developers, and device manufacturers. The use of standard interfaces also facilitates the integration of Audio Object Rendering capabilities into larger multimedia and metaverse systems.
A distinctive feature of CAE-AOR is its support for user-driven interaction with the Audio Scene. The specification enables a user to modify relevant aspects of the listening experience, including the Listener Point of View, the Audio Object Position, the remixing of audio programme elements, the substitution of Audio Objects, the introduction of object motion, and the modification of characteristics of the Acoustic Profile. These capabilities make it possible to create personalised and context-aware audio experiences while preserving interoperability among independently developed implementations.
The CAE-AOR V1.0 specification defines the architecture, AI Modules, and Data Types required to implement Audio Object Rendering functionality. Together, these specifications provide a common basis for the development of interoperable systems capable of transforming Audio Objects and Audio Scene Descriptors into rich, flexible, and interactive audio experiences for a broad range of applications.
In all Chapters and Sections, Terms beginning with a capital letter are defined in Table 1. All MPAI-defined Terms are accessible online. All Chapters are Normative unless they are labelled as Informative.
