At NAB 2019 on the Avid Stage, re-recording mixer for TV Scott Weber, who has mixed shows like Lost, Person Of Interest and Westworld, discussed how to mix TV shows to meet the different US broadcaster’s and Network’s delivery specs and how things vary from network to network and what impact that has on the mix and deliverables.
Scott started his presentation with a trailer from The Tracey Ullman Show that was the first TV show he mixed in the early 90s, mixing it at a music studio using outboard gear like Compellors, Dominators and Exciters, to help to mix the show so it was as loud as the commercials. To achieve this, they would mix a show and then watch in on the air and make adjustments based on how it sounded on the air. This was all before DAWs came along so they were laying their dialog, sound effects and music to a 24 track machine and the levels were much more about music studio workflows like watching VU meters.
From 2004 to 2010, Scott mixed the iconic Lost TV series. Although they were mixing in 5.1 they choose to mainly monitor in stereo because that was what they thought was the best thing to do in order to make sure it was going to translate on-the-air. For Lost, when it came to delivery specs, it was all about peak level. To make sure they were complying with what the broadcaster was using, they visited the engineering team at ABC Prospect and found they were using a special set of RMS Peak meters to check compliance and so Scott and his team persuaded the broadcaster to let them borrow one of these meters to make sure that their mixes would comply with the broadcaster’s delivery spec.
About half way through the Lost series, ABC Prospect changed to the DK Meters, which were much more accurate and could also measure True Peak levels. What this meant is that their mixes no longer met the delivery spec, so Scott and his team got a set of the same DK meters that ABC Prospect were using so that they could be confident that they would be able to deliver compliant mixes.
Next Scott discussed delivering BS1770 based mixes to the ATSC A/85 spec and if you want to know more about loudness compliant mixes then do check out Mike’s excellent tutorial series Understanding Loudness - For Everyone Working In Audio Post Production. Around the world especially in the broadcast audio sector there has been a change taking place from working to peak audio levels and normalising to peak level to measuring and normalising to the new loudness specifications.
In just over 1 hour Mike Thornton helps you understand how the new loudness standards affect you and the clients you work for. Mike has been delivering loudness training courses to broadcasters in the UK, Europe and further afield and with this tutorial series you too can benefit from his expertise in 5 lessons. Mike’s course is for anyone working with the loudness workflows and needs to produce content to be compliant with the loudness delivery specifications like ATSC A/85 for the US and Canada and EBU R128 for Europe. This can be live audio mixers, audio post production editors and mixers, compliance teams, video editors, anyone working in the delivery of broadcast content.
So Scott found himself in this new world of BS1770 and ATSC A/85. He summarised the 2009 edition of the ATSC A/85 spec, which in essence had two specs. One for short-form content, for which the full mix was expected to have an integrated loudness of -24 LKFS and a maximum true peak level of -2dBTP. Then for long-form content the spec was -24 LKFS but measuring the anchor component, which would usually be the dialog, using Dolby’s Dialog Intelligence algorithm, with a maximum true peak level of -2dBTP.
Scott also covered how BS1770 evolved and how gating was added with BS1770-3 so that long periods of low level audio did not skew the overall loudness measurement for a program.
Although all the loudness related delivery specs are based on one universal standard - BS1770, there are differences around the world with standard like EBU R128, OP-59 and ATSC A/85, which each broadcaster and network can interpret. For example, in the UK, we have a common delivery spec for all UK broadcasters from the Digital Production Partnership and the DPP spec is based on the EBU R128 spec, but there are subtle differences.
As Scott outlined, in his presentation at NAB 2019, each of the US broadcasters and networks have interpreted the ATSC A/85 spec somewhat differently.
For ABC, the target loudness for the complete program is -24 LKFS on a full mix with a plus and minus 2 LKFS tolerance with a maximum true peak level of -6dBTP.
For NBC on long-form content, the complete program target loudness for dialog should be an average of -24 LKFS again with a 2LKFS tolerance both sides and a maximum true peak level of -2dBTP.
For Fox they measure the target loudness of each program act or segment to be -24 LKFS on a full mix with a plus and minus 2 LKFS tolerance and a maximum true peak level of -6dBTP.
For Netflix they now have a delivery spec where the dialog loudness should be -27 LKFS using the Dolby Dialog Intelligence algorithm (gated dialog) with a maximum true peak level of -2dBTP. You can learn much more about the Netflix spec in Mike’s excellent articles…
What is interesting is that one of Scott’s comments on the latest Netflix delivery spec is that if you mix a show in a theatrical theatre with a film calibrated monitoring system at 85dB SPL then your dialog comes out at -27 LKFS and Scott added that in his view the Netflix spec is closer to a theatrical spec than it is to a TV delivery spec.
Scott then shared that if you mix at a lower SPL level than the Dolby approved 85dB SPL, then your mix will be louder and you are going to print to your print masters louder and its also going to make you mix a lot less dynamic so things like background sounds and music will end up being a little louder and so for television mixing, as opposed to theatrical mixing, it is actually advantageous to mix at a slightly lower SPL level.
Scott then opened up a Pro Tools session and discussed the challenges of delivering mixes to be compliant with the different US broadcasters and networks, especially what happens to a mix for delivery to Fox who require each part or act to be loudness compliant in its own right.
Whilst he was in his Pro Tools session Scott also showed the loudness metering that he uses and that he still uses a VU meter and a Dorrough meter as well as the Waves WLM Plus meter as he prefers to have legacy meters that he is comfortable with as well as the newer loudness metering. He also acknowledged to being a recent convert to the Nugen Audio VisLM with its comprehensive history.
Scott explained that he uses a mastering stage after completing the mix to help provide the deliverables, fine tuned to each of the broadcaster’s or network’s requirements, especially when it comes to what it takes to deliver a Fox compliant mix.
To conclude his presentation Scott outlined what happens to his carefully crafted mix once it has been delivered to the broadcaster or network. In talking to them the first thing he found out is that they check the loudness of your mix and even though it might be say -23 LKFS, which is within the plus and minus 2 tolerance, they will normalise it to be exactly -24 LKFs, so if you thought you might cheat the system a little bit and deliver a mix that is slightly louder, albeit still compliant, then that will be in vain as the networks will normalise it back to -24 LKFS.
However when Scott heard his mixes on air, he thought he could hear some other things going on that weren’t in the mix as it had left him. To investigate, he hooked up his digital recorder to his cable box and took a couple of samples of programs he had mixed. He also talked to some of the networks and found out that some of the affiliates make changes to programs by reducing the dynamic range or Loudness Range of program mixes if they feel that the mix is too dynamic. He also discovered that the ‘on-demand’ version could be different to the broadcast version as well as different to his original mix. In the screen grab above the top track is the off-air recording Scott made, the middle track in the on-demand version and the bottom track is his original mix and you can see that the dynamics are different for each one of them. So it is clear that certainly in the US the broadcasters and networks are making dynamic range changes to mixes even though they were delivered as being compliant to the respective delivery specs.
Scott’s conclusion was that the closer we deliver our mixes to specs provided by the network, the more transparent it is going to be on the air.