ABSTRACT

Together with nine European partners, b<>com is participating in the ORPHEUS research project, which aims to invent new workflows for producing, broadcasting and playing back object-oriented audio content.

Among the many challenges associated with the adoption of objectoriented audio representations by the public and professionals, is the lack of dedicated sound recording techniques and tools. In particular, capturing sound objects together with their associated metadata, such as the object’s position, remains difficult.

In this paper, we focus on the problem of localising a sound source by comparing the source signal, as recorded using a spot microphone, to the signals recorded by a distant microphone array. We designed an algorithm for estimating the direction and distance of the spot microphone, relative to the main microphone array. The performance of the algorithm is assessed using both simulated and recorded sound signals.

INTRODUCTION

Object-based audio is a relatively new paradigm for representing audio content. Traditionally, audio content such as movie soundtracks or radio programs was stored and distributed in a format that corresponded to the device used for sound reproduction. For instance, the music in a compact disc is in a stereo signal format and is assumed to be played back over two loudspeakers distributed accordingly.

On the contrary, with object-based audio representations contents are decomposed into distinct objects, which typically define sound sources. These sources are characterised by one or more audio signals and some metadata defining its position, width, whether it belongs to the audio foreground or background, etc.

Object-based audio representations are revolutionary in that they allow more immersive and more interactive user experiences. More immersive, because 3D audio content represented as objects can be played back optimally for any speaker configuration or rendered binaurally over headphones. More interactive, because they offer the possibility to displace sound sources or to change their level at the user end.

Rendering can also be adapted to suit the listening conditions. For instance, the dynamic range of the content can be reduced for hearing-impaired listeners, when listening in noisy environments.

The ORPHEUS project is a research project funded by the European Commission under the Horizon 2020 programme. Its goal is to prepare the future of audio broadcasting with a highlight on object-based audio representations.

Gathering broadcasters, researchers and engineers from four European countries (UK, Germany, France and the Netherlands), the project aims to define new tools, workflows and standards for producing, archiving, broadcasting and consuming object-based audio experiences. 

Download the full technical paper below