In audio engineering, whether for broadcast, live sound, games or film, one typically has many different recordings of sound sources (vocals, sound effects, wind instruments, etc.), each one represented on a separate track. These tracks need to be heard simultaneously in the final audio mixture. Each track may have been created in a different way, in a unique environment, at a different loudness level than other tracks. Some tracks may mask each other, some may be too loud or too quiet, some may blend in well with the others most of the time, but then have periods where they sound terrible. The ways in which a sound engineer solves these problems can vary based on the individual, making it a difficult problem to model. Audio production is still very time consuming, and requires a complex set of creative and technical skills. Furthermore, there is a lack of fundamental understanding of how we perceive mixed audio content, and what user preferences might be for different production styles. In this book, we provide insight into these areas of ambiguity, and attempt to provide computational models of these inherently complex subjective processes.
In related domains such as image and video production, a plethora of tools including autofocus, face detection, context-aware editing and red eye removal provide intuitive and simple ways to manipulate and interact with content, however intelligent system in audio production tend to be less prominent in commercially released hardware and software. Commercially available audio editing devices are essentially deaf, they tend not ‘listen’ to the incoming audio, and have no knowledge of the sound scene or of preferred production techniques. Researchers today aim to change that situation, making systems able to ‘listen’ to the audio and suggest or even perform steps to improve mixing and mastering. In this book we detail the state of the art in intelligent music production systems, ranging from systems for semantic control of audio processing tools, to semi- and fully-automated mixing and mastering systems.
This book brings together both published and unpublished work in the field of audio and presents a range of academic research projects in an understandable and usable format. Readers of the book will be provided with a range of demos and examples, including a repository of links, code snippets and software tools for incorporation of the techniques into their workflow.