The Loudness Challenge, Part One: Linear Acoustic Q&A
By Dan Daley
SVG Audio Editor
The issues surrounding relative loudness within broadcast programs, a controversy that has become more heated in the DTV era, continues to be top of mind for sports-broadcasting professionals. Tim Carroll helped birth Dolby’s AC-3, Dolby E, and Dolby Surround products before founding Linear Acoustic, a company that has been heavily focused on broadcast’s loudness issues with products like its AERO.air series of products. Carroll recently discussed the philosophy both he and his company have regarding loudness. Later in the month, we’ll talk to mixers facing these challenges and list some of the product solutions they’re using.
Sports Video Group: How do you define the “loudness” issue? To what extent is it composed of quantitative measurable elements, such as levels, and of more-subjective perception?
Tim Carroll: Until recently, quantifying loudness has been a largely subjective process. First, with the introduction of a practical meter for making a long-term integrated measurement, Dolby set the stage with their LM-100. Next the ITU [International Telecommunication Union], through the BS.1770 standard, has taken input from broadcast and loudness experts and created an international recommendation that all manufacturers can support. The important thing about these measuring systems is that the subjective nature is largely removed and the loudness is reduced to a single number.
More work is being done by the ITU and others to refine this further, but, for the first time, it is possible for two facilities to get the same numeric result representing the loudness of a given program.
However, loudness is only part of the issue. Dynamic range is the other difficult part. A digital delivery system that reaches all the way to consumers coupled with its accompanying metadata has the ability to keep loudness and dynamic range as the two separate issues they actually are. Unfortunately, with all of this new digital room comes some challenges. An evening of programming that is loudness-matched could also be too far dynamic for many viewers.
SVG: In what ways is broadcast sports audio specifically affected by loudness problems?
TC: With proper metering tools, loudness can be quickly judged, and mixers can get back to mixing. Crowd roars and car crashes can be mixed in at levels that talented mixers know will translate well to the typical consumer living room or viewing environment. This is not always how it plays out.
Managing a mix to ensure that it is dynamic enough for the enthusiast but controlled enough for the average viewer is difficult. The audio portion of DTV in the U.S. is designed to allow optimizing a mix for different environments using metadata. This data must be generated properly and encoded along with the audio for delivery to consumers; this is a very large challenge.
All it takes is one program with bad or missing metadata and consumers perceive a loudness problem and complain. It is happening often enough that Congress has decided to get involved.
SVG: Can you suggest an example or two?
TC: The classic example of programming versus commercials [is] an easy target. Interestingly, it is less about matching the loudness of a commercial to the average loudness of a program than it is about matching a commercial to the last 30 seconds of a program.
The program may end quietly for effect, dialogue could be comparatively low, viewers turn their volume controls up to hear the last line of dialogue, and Wham! A commercial that was properly matched in loudness to the “average” of that program is now way too loud because the viewer turned up the volume to hear the dialogue of the program. Not the fault of the commercial, not really the fault of the program, but a result of artistic use of dynamic range.
SVG: Can live mixers “protect” their relative levels from downstream processing? Would they even want to?
TC: In general, the less overall processing done on audio before transmission, the better. Compression effects tend to multiply rather than add, and a highly compressed, very dense program will actually cause processors to turn down the level and will add to loudness inconsistency.
In our view, the transmission processor is a fail-safe, meant to catch outliers. If you hand a processor programs that vary wildly in loudness and expect it to instantly produce smooth audio, it can be done but with sacrificing the quality of the programs that did not really need processing.
It is critical to remember that a processor has to be adjusted to handle the worst case, not the best case. Metadata can help a great deal. Mixers can best protect their mixes by either supplying correct metadata with their programs or matching the standard set of metadata parameters specified by the network they deliver their programs to.
For example, if a network specifies that they want average loudness to be -24, there is no difference between delivering a program with dialnorm measured and set at -24 and delivering that same program with the loudness and the audio measured and set to -24.
SVG: What are Linear Acoustic’s solutions to loudness as an issue for sports-broadcast audio?
TC: We provide ITU BS.1770 loudness-measurement meters and monitors, metadata generators, and processors that support metadata-based processing as well as traditional and permanent PCM processing and anything in between. We strongly believe in helping content producers protect the integrity of their artistic creations and believe that a careful combination of metadata and audio processing result in the most pleasing presentation for consumers.
The tools are available, and we are working with producers and broadcasters and international standards bodies, such as ATSC and EBU, to help protect content and consumers. The solution lies in correct use of audio and metadata and not in legislation.