TV-BAY Magazine

To view this page ensure that Adobe Flash Player version 11.1.0 or greater is installed.

Managing loudness for broadcast TV by Anothony Wilkins O ften viewed as the “ugly sister” when compared to its more glamorous rival video, audio is finally receiving some much deserved attention in broadcast circles. The trigger for this realisation is an increasing shift in awareness that viewers are interested in not just the pictures they are seeing, but equally the sound they are hearing. In particular, there is an increasing amount of attention being paid to how loud broadcast audio is perceived to be. Why perceived? Well unlike peak levels that were always used as a reference, loudness is extremely subjective and cannot be measured as a simple electrical signal. How loud a person considers sound to be depends on many factors including frequency, type of content, listening environment and even the mood they are in. When different program material is normalised to a common peak level, this leads to variances in the perceived loudness level which in turn creates ever increasing complaints to broadcasters. The effect is compounded when there is a mix of content ranging from very wide dynamic range material such as classical music to highly compressed items like promos and commercials. So how is loudness measured? ITU-R BS.1770 provides a method of making a very close estimation by modelling the way in which the human auditory system interprets the relative loudness of sound. The main elements are a high pass filter rolling off from 200Hz and a high frequency “shelf” starting at 1kHz lifting higher frequencies by 4dB. Collectively this is referred to as the “K weighting” curve. After filtering, the relative power over time for each channel (L/R for stereo and L/C/R/ Ls/Rs for 5.1) is summed (Ls and Rs have 1.5dB gain applied and LFE is ignored) which produces a loudness measurement. The unit denoted by the ITU is LKFS (Loudness units K weighted referenced to digital Full Scale). So now that we can measure it, what constitutes “good” loudness? There are a small number of standards or recommendations that broadcasters can use to manage the loudness of their content. Of greatest relevance in Europe is the EBU recommendation R128. At the heart of this are three key elements: 1. Program Loudness 2. Maximum True Peak Level 3. Loudness Range R128 suggests that program loudness should average -23 (+/-1) over an entire program. This is an integrated measurement based on continuously updated samples from the start to the end of the piece. The unit of measure adopted by the EBU is LUFS and is directly interchangeable with LKFS. 1 LU is equal to 1dB. Maximum True Peak level is limited to -1dBtp to prevent over modulation downstream (this is particularly important when a lossy codec will be employed) and requires the use of an oversampling meter to catch the fast transients that a standard PPM could miss. Loudness Range (LRA) does not have a specified “correct” value but one should be aware that a Hollywood feature film mix with an LRA of c.25 could be fine for a user with a high quality home cinema system with large speakers but would not be well suited to reproduction on an average consumer’s flat screen TV with typically very small transducers. So we know how to measure loudness and we have a recommendation of where it should be, what does that mean in practical terms for broadcasters? Although in the UK there is no specific compulsion to conform (unlike the US which has passed the CALM act into law mandating conformance with their standard) there are reasons to start paying attention to loudness. Many European countries have already adopted the practice of normalising to -23 LUFS so this could have implications for International program exchange and the likelihood is that pressure to conform will build at some point. For self-originated content, now is a good time to start mixing to the target program loudness rather than peak level. Jünger Audio offer QC/ loudness monitoring solutions like the M*AP to help achieve this. For content received from other sources, it is good practice to assess, and if required correct, as early as possible in the chain, for example at ingest. Again, Jünger have solutions to perform this function. File based products can be very effective here as they offer “infinite look-ahead” by scanning the entire file first (usually much faster than real time) and then scaling as appropriate. There could always of course be content that slips through and in this case, a Jünger T*AP loudness manager placed before the DVB 60 | TV-BAY MAGAZINE: ISSUE 77 MAY 2013 TV-BAY077MAY13.indd 60 02/05/2013 21:19