Issue link: https://digital.copcomm.com/i/1546135
C I N E M A A U D I O S O C I E T Y. O R G 27 following instructions and didn't have the decision-making power to change anything. Either that or they did actually have some power but didn't have the correct knowledge and were too stubborn to admit it. Yeah, I said it. You know who you are… A few years back, Jim Starzynski, NBC- Universal's chief audio engineer, Chair of ATSC's S42 Audio Group (and longtime sympathetic ear to us mixers), asked me to contribute to an AES paper about the recent rise of subtitle use "AES TD1009—Improving Dialogue Intelligibility in Media." (It's a great report, and if you haven't checked it out, it merits some of your time.) In the course of contributing to this paper, it became clear to me that by leveraging the weight of the CAS, we could lobby the ATSC (being the body that regulates broadcast television in the United States) to update A/85 based on the real-world experiences our members and the mixing community have had since its inception in 2012. At the same time this was going down, Marc Fishman CAS, Jamie Hardt CAS, and a group of our members formed the Industry Technical Standards Committee. The goal was to pool and leverage the knowledge and enthusiasm of CAS members into a brain trust that could do outreach and try to positively influence the industry in technical matters that affect what we do in the real world. Early on, we focused on that all- important concept, the "Anchor Element," which had become almost universally neglected by the broadcast networks and cable channels—and one notable big streamer. The approach that a mix should have its loudness measured according to an audio element that the home viewer will perceive as the consistent "center" of the program makes so much sense that it's been mind-bogglingly frustrating to have to argue for it. For almost all narrative programming, the listener's focus will be the dialogue. For music videos or With studio design legend Russ Berger There is a veritable alphabet soup of standards and specifications pertaining to television sound, and it can be very intimidating—or at least mind-numbing—to mixers just trying to get the show to pass QC. Part of the goal of our committee is to try to demystify these and put them in their appropriate creative context. With that in mind, we've put together a short list of terms and concepts you may run into in an audio spec sheet. Anchor Element — The loudness "reference point" around which the other elements of a program are mixed. If the program is dialogue-driven, the anchor element may be dialogue; if it's a music video or musical program, the anchor element may be music. The idea is that the anchor element is what a viewer would be paying attention to when they set the volume on their television, so if the anchor element is at a consistent loudness, they will never touch their remote. ATSC — The Advanced Television Systems Committee, the body that develops the standards for high-definition television in most of the territories that had previously broadcast using the NTSC television standard: the United States, Mexico, Canada, South Korea, and several countries in the Caribbean (Japan previously used NTSC for television but does not use ATSC). ATSC A/85 — The ATSC recommendation for the loudness of television programming on ATSC television. The ATSC recommends that for a long-form program, the anchor element meters at -24 LKFS, ±2 LU. Short- form programs, such as commercials or interstitials, are measured over their whole program mix, not just the anchor element. CALM Act — The Commercial Advertisement Loudness Mitigation Act of 2010 (Public Law 111– 311), the U.S. federal law that authorizes the FCC to regulate the loudness of television broadcasts. The FCC in enforcing this law endorsed ATSC A/85 as the standard for audio loudness on US television broadcasts. dBTP — Decibels True Peak. The projected maximum signal level of a recording after it has been converted back to analog or to a different sample rate. Dialnorm — A metadata field present in a Dolby AC-3 audio stream that can signal to a decoder the relative level of the dialogue in the program at that time. If the broadcaster sets this field accurately when the program is transmitted, the receiver can adjust its output level to keep dialogue at a consistent level between programs. Dialogue-Anchored — A qualification on a loudness specification, where dialogue is designated as the Anchor Element. Long-form content, according to the new A/85 recommendations, should be dialogue- anchored. Dialogue-Gated — An Integrated loudness measurement that only counts those sections of the mix that contain speech, as detected by Dialogue Intelligence. Dialogue Intelligence — An algorithm developed by Dolby Laboratories for detecting if speech is present in a short sample of audio. Dialogue Intelligence is used by loudness meters to implement their dialogue gating feature, to reject audio that doesn't contain dialogue from the integrated sum. Dialogue Intelligence is a trained machine learning model, with open weights, based on zero-crossing rate, spectral skew, and other audio features. EBU — The European Broadcasters Union, an international organization of national broadcasters, mostly though not entirely in Europe that have created several standards for the exchange of programming A GLOSSARY OF STANDARDS between their members. (Fun trivia: The EBU also operates Eurovision and the Eurovision Song Contest.) EBU R128 — The EBU loudness recommendation among its member broadcasters. The EBU recommends, broadly, that a program to be mixed to an integrated level of -23 LKFS, ±0.5 LU, although it does discuss anchor elements and makes recommendations thereof. ITU — The International Telecommunication Union, a United Nations agency and successor to the International Telegraph Union that drafts voluntary standards over a broad range of communications technologies, including radio bands, television, satellite orbits, Internet protocols, and other things. The ITU publishes several standards related to the measurement of audio that ATSC and EBU standards rely upon. ITU BS.1770 — The mathematical definition of an LKFS meter, as used by ATSC A/85 and EBU R128 to calculate LKFS, True Peak levels, LRA, and other measures. Any meter that has the three bars of momentary, short-term, and integrated loudness, such as the Nugen VisLM, Youlean Loudness Meter, Avid Pro Limiter, etc., is a BS.1770 meter. BS.1770 has multiple annexes ("-1," "3," etc.) that qualify the behavior of the meter and define how to measure wide multichannel formats and Dolby Atmos. LFE material doesn't count against the measurement, while surround channels are weighted at 1.41 to 1. K-Weighting — A weighted spectrum used by BS.1770 to measure the perceived loudness of a program. K-weighing applies a high-pass filter with a corner around 40 Hz, and a high shelf with 4 dB of gain at a center frequency around 1,500 Hz. LRA — Loudness Range. The difference between the loudest and quietest short-range measurements (in dB), after having discarded the quietest 10% and loudest 5%. ATSC A/85 does not recommend a loudness range, though some platforms maintain a voluntary standard as part of their specifications. LU — Loudness Units. The unit of measuring program loudness defined by BS.1770. LU's are meant to be a subjective measure, reflecting how a listener experiences the loudness of a program or the subjective jump between a louder and quieter program, and is based on empirical studies. LU's are logarithmic over signal levels (an LU is equal to one decibel, after weighting and gating), and relative. LUFS — Loudness Units, Relative to Full Scale. A measure of the loudness of a program in LU's relative to full scale. See also: LKFS — Loudness, K-Weighted, Relative to Full Scale. BS.1770 Annex 2 tweaked the definition of the meter's pre-emphasis filter and named it the "K-filter," and accordingly renamed the relative measure "LKFS" to make this change clear. "LUFS" and "LKFS" are identical in modern usage. Momentary Loudness — The integrated loudness, in LKFS, of the previous 400 milliseconds of audio. In a running program, momentary loudness is calculated every 100 milliseconds (each window has a 75% overlap with the previous window). Per Segment or Per Act — A requirement, in some broadcaster specs, that each segment or act of a program independently have a complying integrated level. This is, we can accurately state, a folly. Short-Term Loudness — The integrated loudness, in LKFS, of the previous three seconds of audio.