The seemingly simple statement that measuring dialog loudness is straightforward depends fully on understanding that there are only certain valid parts of dialog that are appropriate candidates for being measured. The dialnorm system in Dolby Digital (AC-3) as standardized, relies upon the measurement of so-called "normal spoken dialog" and does not include whispering, shouting, or singing. The Dolby LM100 Broadcast Loudness Meter follows these rules with its Dialog Intelligence feature, starting the loudness measurement when valid dialog is detected, pausing it when valid dialog is not detected, then continuing when valid dialog returns. If an actor or actress starts screaming, whispering or singing, dialog measurement pauses--but the possible annoyance resulting from quiet content being overwhelmed by screaming increases.
What this means is that there is an even more narrow range of program material that can be measured with today's tools. As it is arguably normal spoken dialog that humans seem to use as an anchor for judging loudness, how can systems ensure the correct matching of content that does not contain normal dialog? They simply cannot do it without additional help. Assuming that normal spoken dialog is the reference, then all other parts of the programming must be a comfortable distance away from this reference. This can be done manually with artistic intent, hopefully tempered by the needs of the typical viewer, or it can be done automatically.
CATEGORIES
