(l-r) Jim Starzynski, NBC Universal, J. Patrick Waddell,
Harmonic Inc., chairman Tom Scott, Onstream Media,
Kenneth Hunold, Dolby Laboratories, and Kent Terry,
Dolby Laboratories. Photo: Mel Lambert
By Mel Lambert | content-creators.com
Los Angeles (November 2, 2009)–The recent SMPTE Annual Technical Conference & Expo included a fascinating session that addressed two critically important aspects affecting both content providers and broadcasters. Hosted by Tom Scott from Onstream Media, Audio Issues in the Digital Environment considered lip-sync errors that are produced during just about every stage of production, post and delivery to air/cable, while a separate discussion focused on how broadcasters wrestle with consistent loudness levels for the consumer. The SMPTE Conference & Expo was held at the Hollywood Renaissance Hotel from October 27 thru 29, following a specially organized one-day Seminar on Advanced Media Workflows and MXF.
J. Patrick Waddell from Harmonic Inc. reviewed some of the sources of the lip-sync anomalies, as well as current work towards its resolution. Acknowledging that ITU BT.1359-1 – the only currently available international standard that quantifies acceptable lip-sync tolerances – was formulated by expert viewers using older-generation CRTs and standard-definition images, Waddell explained that new studies will involve HD sources and large-screen plasma/LCD displays. As he pointed out, “The new CEB-20 [published in July 2009 by the Consumer Electronics Association/CEA] provides detailed guidance to manufacturers of consumer set-top boxes and professional decoders with techniques for implementing tighter synchronization schemes. It provides guidance for real-world implementations,” Waddell stressed. A SMPTE Study Group is also currently considering ways of alleviating the problem within production and post-production facilities.
Kent Terry from Dolby Laboratories described a remarkable finger-printing technique currently being developed that looks at video and audio content and generates a low data rate AV synchronization signature. This signal could be carried in a metadata packet, for example, and used to automatically correct downstream sync errors against a previously defined start point. Terry provided real-time examples from Dolby’s internal research project that demonstrated accurate re-synchronization of widely disparate signals; an accuracy of +/- 10 mS is the aim. Other uses for this 3-4 kbps synchronization signature are envisioned, including content identification and verification, in addition to quality control and dynamic metadata – using timed flags to initiate on-screen IDs, and the like.
Waddell returned to review the current loudness dilemma facing broadcasters – centered around the pending Congressional Bill initiated by Anna Eshoo (D-Cal) in early 2008 as a reaction to aggressively loud TV commercials, and now designated H.R.1084, The Commercial Advertisement Loudness Mitigation Act (CALM) – before outlining activities of the Advanced Television Systems Committee that had been initiated several years prior to Eshoo’s bill. ATSC’s Technology and Standards Group’s S6 has been looking at both loudness and sync anomalies; the former under TSG S6-3 and the latter under S6-4.
After drafting for a year, S6-3 has released a document TSG-850r0. Following a membership ballot it is expected that this document will become a forthcoming Recommended Practice ATSC A/85 – Techniques for Establishing and Maintaining Audio Loudness for Digital Television, which describes both methods for accurately monitoring multichannel mixes during production, post and QC, in addition to loudness levels at which materials should be delivered to home viewers. “The membership ballot closes on November 4,” Waddell pointed out, “the same date as ATSC’s Audio Seminar in Washington, DC.” All major TV networks participated in the formulation of ATSC’s proposed RP.
“Whether this RP will forestall the Eshoo bill remains to be seen,” Scott remarked, “but at least now H.R.1084 will reference a real Recommended Practice.”
The chair of ATSC’s TSG S6-3, Jim Starzynski from NBC-Universal, described the contents of the proposed Recommended Practice, which runs to some 67 pages and contains a pair of Quick Reference Guides for TV stations and content providers, and then outlined how NBCU manages cable and broadcast audio loudness by applying the RP’s concepts. “Even though the problem could be at the local cable end, for example, the blame [for loudness jumps] is placed on the entire program chain, back up to the content providers,” Starzynski stated.
NBCU’s Content Delivery Program Specification, finalized in July 2009, lists the use of a broadcast loudness meter to ITU-R Recommendation BS.1770 and setting loudness levels to -24 LKFS, +/- 2 dB, when no metadata is supplied with the content. The dialnorm metadata parameter specified by AC-3 data-compression and multiplexing is then matched to this loudness level. Now any variations in the transmission path will be adjusted automatically within set-top boxes and domestic receivers using conventional dialnorm scaling. “Our strategy uses a combination of four elements,” Starzynski said. “[These include] a content-delivery specification, real-time targeted mixing, processed loudness targeting and loudness scaling.” To ensure that loudness settings match the network’s recommended-24 LKFS level, “cable channels use Linear Acoustic AEROMAX, while the network uses only the DRC contained in the AC-3 bitstream.” Currently under evaluation are two different scalers at NBCU’s TV stations and network for adjusting levels of ingested program material from content providers that do not match this tagged loudness level. “Once new workflows are established,” Starzynski concluded, “the plan is to go to air with them.”
Kenneth Hunold from Dolby Laboratories completed the technical session with a close look at the levels currently being broadcast by cable outlets and over-the-air broadcasters. “Given that, typically, TV programs are delivered without a specification,” Hunold offered, “loudness can vary significantly. Although some networks have asked for programs to be delivered at a specified loudness, have program producers been meeting those specs?” Using loudness measurements in compliance with the ITU-R BS.1770 standard taken from some 80 prime-time TV programs, Hunold showed wide variations both within programming from a target network, as well as between networks and other suppliers. Results indicate that program loudness varies from -30 to -21 LKFS, and also loudness values that did not match the stated dialnorm value. Hunold discussed techniques for managing program loudness, utilizing methodologies contained within the proposed ATSC RP, and which would overcome the all too often consumer litany of “Change the Channel; Change the Volume.”