Audio Loudness for Broadcast – AES 2010
November 4, 2010 – San Francisco, CA – A discussion panel was held at the 129th AES regarding the new standards for loudness balance for TV broadcast and commercials. The panel was led off by Pat Waddell of Harmonix who brought the audience up to speed with the what were the governing documents, and the specification for “loudness balancing”. The main governing documents are the ITU-R BS.1770-1 loudness measurement specification and ATSC AES 85:2009 recommended practice for audio loudness specification. The key issues revolve around digital TV transmission being able to address a 100db dynamic range for audio. This is about 2X the dynamic range that was possible with analog audio broadcast.
To insure that there is no abuse of the dynamic range, and the viewer does not have to keep adjusting the volume between shows and commercials. Both US and European television have come up with specifications on how the loudness balance should be measured and best practices for implementation. NBC Universal presented how they were addressing this issue from a content and a full commercial operation perspective. On thier network owned stations and common backbone, they have a full digital broadcast system which handles content that is 90% digital file based system. The file base strcuture give the ability to normalize the audio to a specific level per the AC3 specification standard. As they have a closed loop environment which meters the full signal flow and then uses an ATSC AC2 encoder at the transmitter to insure that the loudness is fixed at -24LKFS.
The representive of the Cox affilated stations identified a different problem. As the stations in their network are independently owned and operated, and some are minimally staffed, a full front to back digital system with in-line metering is not possible. Local stations get a combination of file based data, and local commercials/ads that are video tape based. This is mixing the dynamic range models and in a minimally staffed station, they do not have the bandwidth to both manage the transmission feed and adjust the dials based on meter response and stage/channel the content into the transmission pipeline. In these cases, unless the data on the tapes is transcoded to AC3 format with appropriate metadata, the playout will require a manual control of the volume to meet the specification.
There was a lot of discussion on metadata, data models, and algorithms for balancing the loudness, but they were all based on having the audio data in digital format. The discussion was based on mono and stereo audio, but the specifications are extensable to 5.1 sound. At this time, there is no broadcast standards for 7.1 sound. As a scope of the issue, the loudness specificaition is targeted at a per channel basis for peaking and balance, chanel to channel is not covered. Furthermore, for new “smart” or “connected” TV, there is no specification standard for playback on the OTT content that is collected from the web or other streaming sources. The interaction of sound from apps that may be implemented in the context of broadcast signals is also not addressed in the specification or in the discussion.
