Livening up surround-sound in digital television

White Paper

Livening up surround-sound in digital television a technical white paper by Jonny McClintock, APTX www.aptx.com
1 | © APT Licensing Ltd., 2009 White Paper
INTRODUCTION One of the many advantages of HDTV is the ability to broadcast multi-channel audio and allow listeners to immerse themselves in a true surround-sound experience. However, it would be true to say that the audio portion in HDTV has been the poor relation when compared to the video component. Further development in video is slowing and suffering from the laws of diminishing returns with regards to engineering input versus return on investment. In contrast, the audio component is virtually a virgin territory as can be witnessed by the limited number of options currently available for broadcast, acquisition, production and processing. Today, home theatres can decode surround-sound in six channels, i.e. 5.1 (Left, Right, Centre, Left Surround, Right Surround and Low Frequency Effect/Sub Woofer) and broadcasters will typically process these six channels together with an additional two for Left / Right Stereo. However, the pressure for greater channel density is set to increase as 6.1 and 7.1 (Dolby Digital +) become more prevalent (research in Japan [1] is also looking at 10.2 and 22.2), as multi-lingual communities seek to offer more services in multiple languages and as the requirement grows for additional audio channels for ancillary services such as "Audio Description" (AD) for sight-restricted viewers [2]. The companies that offer solutions for broadcast, namely Dolby, DTS, and SRS will continue to innovate, and industry bodies will continue to standardize to ensure that content is delivered into the home in a responsible and balanced fashion. However, the focus of this paper is to look more closely at the workflow through the production and contribution phases and to examine the challenges for facilities addressing the issue of greater channel density. DOBLY E The video signal has an embedded AES3 stream which can run a Left / Right stereo of linear audio but a typical HD video signal will have a 5.1 mix, a stereo or matrix stereo, descriptive video, and secondary audio programme which equates to 10 channels of audio. To cope with this requirement broadcast facilities usually run a compressed audio signal, typically Dolby E, on this AES3 stream to increase the density from 2 channels to 8 channels (5.1 plus a stereo pair). If there is a requirement for any additional channels, an additional Dolby E stream is sent on another AES3 stream. However this doubles the effort in additional D-to-A's and requires extra levels of routing and wiring. In effect a highly laborious, time-consuming and expensive solution. This paper will present the case that there are a number of frailties associated with Dolby E, the most notable of which are the latency, audio quality and general awkwardness in terms of operation and implementation. Despite these shortcomings, Dolby E technology has been widely adopted for a number of reasons. Firstly, no other supplier stepped up to the plate to offer broadcast facilities a truly alternative solution and secondly, Dolby concentrated on one issue, i.e. metadata and went into marketing overdrive to ram home the benefits of this feature.
2 | © APT Licensing Ltd., 2009 White Paper
If some time is taken to explore the weaknesses in Dolby E, this may lay the grounds for a solution that will actually meet the needs of the industry. DISSECTING DOLBY E Let's start with a look at latency. The delay through Dolby E is 40 - 66 milliseconds, which varies depending on video frame size. The resultant of this long latency is a full video frame; as such an entire cottage industry has been spawned with the aim of addressing and managing this one issue. Regarding audio quality, Dolby E is fundamentally based on psychoacoustic masking or perceptual coding principles (a highly processing intensive technique which causes the latency). For the purpose of clarification, the concept behind this principle is quite simple: take two signals side by side in the time axis, if one signal has a higher amplitude than the other, a mathematically generated ear estimates if the lower amplitude signal can be heard or not. On the assumption that the signal cannot be heard it is considered irrelevant and discarded. In this way, broadcasters and listeners are being dic... [download for more]