This post should serve as a little guide to the resources available on-line on the topic of audio levels in different media. It has been compiled due to the big frequency of questions on the topic, and thanks to the big amount of answers in this forum!
Since audio in media is an ever-changing field, this post will be updated as I stumble upon new and interesting infos or links. If you have insight into data that you think should be included or corrected, please PM me, or post it here.
For further, specific questions on mixing levels, you can post in this thread, or start a new one.

DISCLAIMER: if you're coming from the MUSIC MIXING/MASTERING BACKGROUND, read this thread first (posts 1-5): GENERAL Average DB settings?

Movie theater

There are no guidelines in terms of average loudness, peak or any other level measurement. You achieve proper levels by properly calibrating your listening environment, so that it resembles the environment of the theater.

To calibrate your room, read this:
DUC: Room Calibration for Film and TV Post
(or, in a nutshell) (or, a tutorial video) (or, as a last resort, if you can't get an SPL meter)
Then mix by ear. "If it sounds good, it is good" - JoeMeek.

Here's a useful discussion:
https://gearspace.com/board/post-pro...st-levels.html

However, there is a maximum loudness level for theatrical trailers and commercials which is measured with the Dolby Model 737 Soundtrack Loudness Meter.
Trailer loudness should not exceed 85 dB Leq(m), as regulated by TASA.
Commercial loudness should not exceed 82 dB Leq(m), as regulated by SAWA.


DVD

Here, same rules apply as with the theatrical mix, except that the monitoring is different (near-field, no X-curve), the room is smaller, it is calibrated lower, AND there is the dialnorm parameter if your sound is AC3 encoded.

Read about dialnorm here:
https://gearspace.com/board/1112798-post7.html (section about Dialogue Level)
and here:
Home Theater Hi-Fi: Dialogue Normalization

You have to determine your target dialnorm BEFORE you start mixing, so you can adjust your listening level accordingly. Most DVD's are mixed for dialnorm -27dB (because that setting is the most compatible with the theatrical mix), but some use the full dynamic range (-31dB).


TV (everything BUT commercials)

Every broadcaster has its own specs. You have to get the specs of your target TV channel.


Detailed Specs

They can be very detailed, like the Discovery specs or the PBS specs (section 3). They will tell you exactly what is your max peak level, average dialogue level, average overall level, what measurement instrument is to be used etc. Meter that the US networks usually specify is Dolby LM100 (*this sentence needs an update, as it seems a lot of US mixers are using different software meters these days*)

As of late 2010, European broadcasters are beginning to encourage the EBU R 128 recommendation, which stands a chance of becoming a de-facto standard for Europe. France and some other countries have started using R 128 since Jan 1st 2012.
Check out this great document by Florian Camerer of ORF and PLOUD: On the way to LOUDNESS nirvana

Here are two threads about mixing against LM100:
Mixing with the Dolby LM100
Anyone have experience mixing while adhering to specs monitored by the Dolby LM 100?

This is great! A post by Mark Edmondson, Audio Post Production Supervisor at Discovery:
Dolby LM100 and Discovery deliverables - Digi User Conference


Basic Specs

The other extreme is on the minimalistic side, like the RTL or BBC specs which give you only the maximum peak level, and the reference level. This is what it's like in most of the Europe, except for the countries where R 128 was adopted.

- REFERENCE LEVEL - it is used for equipment alignment, and doesn't have a direct relation to actual mixing levels.
In EBU countries it is -18dBFS and corresponds to electrical level of 0dBu (per EBU R68).
In SMPTE countries it is -20dBFS and corresponds to electrical level of +4dBu (per SMPTE RP155).
Sometimes refered to as: Zero level, Line-up level, 0VU.
Broadcast Audio Operating Levels for Sound Engineers
Reference Levels on Common Metering Scales
The Ins and Outs Of (Sound on Sound)

- MAXIMUM PEAK LEVEL - this is where you set your brickwall limiter on the master buss, or otherwise not go over it (although in some of the specs, short peaks of 3 to 5 dB over this value are allowed - go figure!). Usually -9dBFS or -10dBFS in the continental Europe, PPM6 in the UK and Ireland.

What can make the confusion here is that the average dialogue level is not exactly specified.

In a perfect world, you would calibrate your listening environment to the ITU-R BS.775-1 standard (-20dBFS pink @79dB SPL/C/slow), [or EBU 3276 and EBU 3276-S if you are in Europe] and then mix by ear. In that case you would get average dialogue levels at around -27dBFS RMS.

However, this way, your mix could turn out too quiet, as there's a loudness war in broadcasting, probably in part due to the loudness of commercials and the loudness war in music.
Average dialogue loudness that works for me (dramatic program, regional stations in the Balkan peninsula) is -23dBFS RMS. To achieve that, I calibrate my monitoring to 75dB, and thus reduce the headroom by 4dB when compared to the ITU's 79dB reference.

However, your best bet is to talk to someone who regularly delivers for the given broadcaster or in a given market, and ask him about his average dialogue level, or how his listening is calibrated. Chances are someone at this forum will be able to help, too.


Further Reading

More about broadcast delivery specs:
https://gearspace.com/board/1181448-post23.html

A great intro to broadcast audio:
Audio for Digital Television

A compilation of links about Loudness in Broadcast:
Loudness Links repository

A seminar on the topic and more:
CAS Seminars - 'What Happened to My Mix?' - The Work Flow From Production Through Post Production - Cinema Audio Society

Dialnorm was to be implemented in broadcast too (as Dolby imagined), but it isn't, so far (and likely won't be):
DTV Audio: Understanding Dialnorm

Food for thought on setting up variable monitoring level:
Bob Katz - Level Practices


TV commercials

Again, you have to get the specs of your target TV channel, but you will most likely only use the max peak value they provide. Below that, you can compress as much as you wish - it's a loudness war, similar to the one in popular music production. This is changing fast though, and now some US broadcasters and some European countries have proper loudness specs - here's a useful thread:
Loudness specs regarding TV commercials In EU countries

These are some of the legal efforts on loudness regulation of commercials:
US: H.R. 6209: Commercial Advertisement Loudness Mitigation Act (GovTrack.us)
UK: UK commercials for TV - perceived loudness issue - Digi User Conference
Congress getting involved in television advertising levels.


Internet / Youtube / WWW

No rules, but most people seem to be allowing the peaks up to -1 or -2 dBFS or dBTP, while the average loudnes is between -15 and -18 LUFS or LKFS. Read all about it here:
Loudness levels for internet teaser-trailer


Radio

Nothing much to say here:
Radio Broadcast Levels

A BBC technical specification:
BBC Radio Resources // Programme Delivery // Glossary

Less is more (straight from the horse's mouth) - Bob Orban and Frank Foti talk about what goes on with your mix in the radio station:
Radio Ready: The Truth


Games

Absence of standards:
Video Game Reference Level/Standards
THX: Establishing a Reference Playback Level for Video Games

A thread at SDO with some advice and some official information from Sony and Microsoft:
Niveau Sonore en jeux vidéo :: Sound Designers.Org (Babelfish English translation)
(Note: the Xbox360 document is in English)

Stickie this immediately! Well done!

Philip Perkins

DVD

Here, same rules apply as with the theatrical mix, except that the room is calibrated lower (see the DUC sticky)

Just one question on this.. I've read that DUC sticky a few times and don't see a definitive answer. Is there an actual lower ref level when mixing a movie for dvd, or is the lower level mentioned due to smaller rooms? Other than cleaning up some of the HF junk that the curve/screen hides, Are there significant level changes to the same film when it goes to dvd? I've always wanted to know this.

J

Here's a thread where Tomlinson Holman himself explains this:
Mixing levels for Dolby Digital: DVD vs. theatrical - The Digital Video Information Network

So, yes - it is only because of the room size/monitoring. Theatrical printmaster can thus be printed directly to DVD(*), dialnorm would be set to -27, so the player itself would trim the playback by 4dB - that is how your 85dB calibrated room playback becomes 81dB. When I mix for DVD, my room is, in fact, calibrated to 85dB - I set my limiter to -4dBFS, and mix by ear, so my dialog ends up about -31dBFS, instead of -27dBFS. Then, when exporting, I add 4dB gain after the limiter, so the dialog ends up at -27dBFS, while the peaks end up just below 0dBFS. I would get the same if I calibrated my room to 81dB, but then I couldn't watch DVDs - the player would duck 4dB further, so I'd be listening at 77dB.
But 81dB is not a good volume for every room, so the best way to calibrate the room (IMHO), is to play many reference DVDs (movies), and try to set a common listening level that will suit them all. Then you can check what level is that to know where you're at (I ended up at 79dB, because I'm in a very small room), and to be able to maintain that calibration.

(*) The re-mixing of theatrical releases for DVD does not happen because of different reference, or DVDs dynamic range - it happens because of:
1) the x-curve, as you suggested;
2) because of different target listening environment - home is not as quiet as the theater, sound is often reproduced from poor speakers, so this re-mix serves mostly for the quieter parts to be raised a bit;
3) the specifics of near-field listening - some errors in details and imaging can get revealed. Also, loud parts can get too loud when listening close.

That said, I will update the DVD section a bit, so that it states the differences in room and monitoring, not just calibration.

Excellent work!! Thank you, Danijel.

Thanks guys! I'm thinking about adding a short intro about general relation between dialog level and headroom in different delivery formats. That would be best if accompanied by a graph.

Also, if anyone can give an intro on levels in radio broadcast, that would be cool to include, too.

Thanks tons for the dvd info Danijel.

One other question regarding mixing for dvd. Do you apply any sort of curve to your room or do you monitor flat? DVD playback in my room always felt too bright so I've applied a 1db per 1/3 roll off starting at 8k and at the other end starting at 63. Do you do anything like that?

thanks again
J

Hey Jesse, so good that you brought this up. I was often wondering why every third (OK, maybe I'm over-blowing this) Hollywood-made DVD has sibilance issues in my monitoring, so I searched a bit, and it turns out that this issue, as well as your issue with HF are related to X-curve, and that Sony setup guidelines for DVD printmaster re-mixing actually do employ a roll-off that is similar to yours (albeit it is applied on the Gennies).

Here's the article:

Mastering for DVD - mastering DVD projects for home theater systems


This excerpt discusses the origin of the problem:
While the frequency response of home theater speakers is all over the map, it's rarely as attenuated on the high end as the X-curve. So a mix made to sound good on an X-curve playback system — in theaters and dubbing stages — is unlikely to sound good on a home theater setup. “It can be really bright in somebody's home,” Vessa says. “And we're not talking about that nice, silky high end that's way out there. We're talking about something in the harsh register, 4kHz to 8kHz. Dialog can be in there, sibilances can be in there, and effects, especially metallic-type effects, can trash your ears in there.”


Here's an excerpt from "The Sony Setup" Sidebar in that article:
Regarding the speakers employed, SPE's spec requires Genelec 1031A self-powered speakers and a Genelec 1094A self-powered subwoofer, or similar. The 1031As are to be used with the -2dB roll-off switch engaged on the tweeter only, with all other switches in flat position. “The -2dB switch,” says Sony's Brian Vessa, “puts a gentle shelf on the top. It doesn't let the response just go wide open out to the sky, because you would never have that at home.”
Skywalker's Jerry Steckling agrees. “There's no such thing as flat,” he says, “except maybe in Nebraska. If there's a speaker that measures flat at 1 meter in an anechoic chamber, when you set it up at 8 feet in a typical living room with a little absorption and all that sort of stuff, it's not flat anymore. The -2dB EQ on the Genelecs does a shelf that starts to roll off at about 6 kHz, and we believe that this is what is happening at home. Our own approach is to leave that switch in flat and apply a nice, smooth curve. But we're within spitting distance of Sony's spec on actual high-end response. We just arrive there a little bit differently.”

Another article right on the topic:
Learning from History - Cinema Sound and EQ Curves
(scroll down to section: What does all this mean for us at home?)

This whole TV spec business just still confuses me. I mix everything at -20 RMS with no peaks over -10dB. Sounds OK, but I hate the limitation. So I just get The Dark Knight domestic TV trailer on DIGI. It sounds great but its peak level is -4dB with louder average levels at -15dB. I mean, what gives?

What's a DIGI? Digital BETA, or something to do with Digidesign
And what are 'louder average levels'? Levels of dialog, or the whole mix with music and everything? Have you seen this trailer on TV after broadcast as well, and how does it sound (in level) compared to other stuff on the same channel?

Yes, Digibeta. ;-) The averages are for the stereo mix. I don't think I've seen this one on TV yet, but I'm just guessing by the caliber of the mix and the enormity of the film that they knew what they were doing, level-wise. I'm just trying to milk some more dynamic range out of my spots. ;-)

It seems this is not correct.
I found an article by TIM CARROLL (Linear Acoustic) & JEFFREY RIEDMILLER (Dolby Laboratories) called 'Audio for Digital Television':
http://www.linearacoustic.com/pdf/NAB-05-18.pdf

On page 319, title 'Reference/Line-Up Levels', the last paragraph concludes with:
"This confirms the concept that the reference/line-up level is NOT the same as the dialog level of a program".

I apologize for any confusion. I'll try to sort this one out, any help or link is appreciated.

More precisely, the question is: if the spec doesn't give explicit value for average dialog loudness (like Discovery does), but only max. peak and reference/alignment level (like BBC, RTL or any given spec in my region), what do you do with dialog level? Do you just rely on 79dB calibrated monitoring, and mix by ear?

Here's one for you...

A certain client sent three recent shows I did back due to peak levels in the Dolby mix. These shows were done to the same specs I have worked with for at least 1 1/2 years. In that time I've delivered at least 60 shows and NONE have come back for this. In fact, in the past 5 years I probably have a 98% passing rate for QCs.

In reading their specs the peak limit is -3db for the 5.1 mix, while their stereo mix is -10db. (My peaks in the 5.1 mix hit -5db on occasion but most were -8 to -6db.) However, in a paragraph earlier in the specs it says peaks should not exceed 8 to 10db above reference level (0 or -20).

So, which level do you meet? How far do you take the argument that it was never a problem before? I'm baffled.

I have the tapes, and will have them laid back to meet the -10 requirement that they want, but I feel that it is going to negatively affect my mix. There's a fine line here between standing up for myself and giving the client what they want.

I can offer this on Radio & TV mixes...our agency produces about a hundred spots per month, which are all sent out digitally to stations all over the country.
Most of the TV is Fast Channel delivery and radio is MP3 via email. Our TV spots go with a -10 to meet Fast Channel's requirements. Radio spots are mixed at the hottest level the mixer will allow, peaking at about +4. Plus we run the final mix thru a maximizer to get it as hot as possible. BTW: some radio stations have a 2mb limit on audio deliveries so we have to keep the rate at around 128. We have one station in smalltown Okla that has a 1mb limit, which requires the spot to be converted at a much lower rate...go figure...audio is their business and they prefer the worst quality they can live with just because they won't beef up their server....anyway....Once the radio spots get to the stations they are usally limited/compressed at least twice before they get to the transmitter. In the case of the last Clearchannel station I worked at we used an Optimod in the master engineering bay, plus the production studios had L/C's for processing all spots. So it got smashed twice before getting up the stick. There are some stations that use a final L/C at the transmitter too, which would process the audio 3 times before it gets to your radio...so it's important to produce at the agency level with as little L/C'ing as possible, otherewise, by the time it gets on the air, it's been squashed into oblivion...Fuel for thought....now you know why the music on FM radio sounds so horrible...

Thank you for the info! Good to hear from someone who's been on both the agency and the broadcast side.

I didn't quite get this..... It seems that this part of your post describes the process at the agency, but later you say:
I suppose L/C means limiting/compression, so it looks contradictory with the statement I quoted first.......

----------

I've heard this before, that it's better not to compress the mix for radio play, and that music that is actually less compressed will sound louder, but then, I've also heard that's not true

Makes you wonder about the current trend to master music releases with so little dynamics?

Sorry, to clarify: I work for a Dallas ad agency that has an in house audio and video production suite. We produce on Pro Tools HD and Avid.
I am the executive producer here and in charge of all audio for radio, tv, web, etc. We use as little Limiting and Compressing (L/C) as possible in the first stage of production where we are building the audio tracks with VO, Music, SFX. TV audio mixes go to Avid at full mix level, but are then limited to -10 in Avid to meet Fast Channel requirements. If we don't send them out at -10, Fast Channel will tweak them down on their end. We'd rather maintain the level control from our end. For radio spot mixes I, again, use as little L/C as needed to iron out detailed nuances of the tracks, and almost always, that will be applied to the VO track. Once the mix is ready to bounce I run it thru a maximizer to crank up the volume of the spot without increasing the gain. We did not use to use the maximizer but after many airchecks of our spots revealed that the other spots in the stopsets were louder than ours I began using the maximizer ( on the suggestion of a fellow producer who had been maximizing his radio mixes for some time ). The result is that our spots air with a consistent volume to other spots in the stopsets at the radio stations. Prior to this position (7 yrs here) I spent 30 years in radio, 27 of that in major market radio. I started out as an engineer for a couple of years, then spent 10 yrs on the air, and the rest as a production manager producing imaging and commercials. That's where the info on the multiple squeezing thru the Optimod at the radio station comes from. I used to drive the freeways with the Program Director, listening to our station, comparing to our competition, with our chief engineer on the cell phone, directing his tweaks to the Optimod, so we could dial in the sound of the station, overcome road noise and gain up the signal.

DVD/BlueRay Encoding + Room calibration needs to be understood as well.
A theatrical title is mixed with LCR @ 85, Surrounds @ 82.
Home theater titles need the surrounds dropped by 3db if mixed in this environment. This can usually be done in the Dolby encoders. Home theater setups are made to be 85 all around.

Update for broken DUC link in Danijel's original post:

NEW UPDATED Room Calibration for Film and TV Post - Digi User Conference


Ken Showler
Audio Post Engineer
Berkley, MI
Forest Post Productions | 248.855.4333

Thanks Ken, I updated the link - it must have been broken for months now.

Not much talk on the forum lately (cold weather?), so I thought I'd use the opportunity to announce that I have (over this past month) updated a lot of stuff in the OP (especially in the TV section).
I have added a lot of interesting new links, and there is the new Games section.

Great links!! So helpful!

Has anyone heard about what spec changes may occur for USA television, as we make the transition to digital broadcast?

Maybe a bit of a weird question, but just for my interest: A typical DVD release, what level is the overall RMS? If you would just mute your monitoring, put on a DVD mix: what is your meter saying, RMS and peak wise? What is the dynamic range?

It would be quite different for 'Fast and Furious' vs 'Last Days'
I think it's not worth measuring really, but if you do it, please post back the results - it would be a fun piece of trivia

I should actually do it to see the results.

The reason I was asking: I mostly mix for TV, but I had to mix some small DVD features (some simple documentary style peices) now and then, in a room that was actually not calibrated correctely spl wise. In those cases, I could use some guidelines to see if my DVD mix isn't abnormally loud or quiet compared to other DVD realeses.

If I'd have to work in different rooms, I'd make a clip with some dialogue lines from reference movies or TV shows (or docs, in your case) and play them briefly to set the monitoring level before I start to work. If you get the dialogue levels right, everything else will be fine. Don't worry about RMS, it depends on the source and the goal/mood too much - e.g. an ambiance can be easily mixed 10dB lower then that same ambiance in another scene.

Hello All,

I've recently join this forum and have been reading a lot of threads, looking for something around this topic. I work primarily mixing independent features that get some theatrical play and mostly DVD/Blu Ray, so I've been doing these number crunches for a long time. There are many questions I have for other mixers here regarding this, but I'll just start with a couple.

1. Has anyone else noticed that on Blu Ray releases, a PCM track is OVERALL 4-5db hotter than Dolby True HD & DTS-HD? I'm not sure as to why this is...but I did some calibration and testing in my room, which is also running a reference Blu Ray player w/ analog outs going into a -10 to +4 line shifter. When authoring a pink noise reference DVD using Dolby and PCM, the PCM imatches the SPL levels that my DAW is cal'd to, but the Dolby playback is 5db quieter in outputing SPL levels.

This seems odd, since many Blu Ray releases with PCM Multitracks will also have a DD track, thus having two mixes at different output levels, and obviously differering for consumers who play a Blu Rays of different audio formats, without changing their volume level.

2. To anyone that is mixing for DVD/BD release, are you only mixing using Bass Management or rather LFE mixing as well? If you do mix with Bass Management, how do you calibrate your subwoofer to compensate for the added low end from the mains?

From my experience, it seems that most commercial DVD/BD releases that get a "remix" before home release, have a very large use of the LFE channel, pretty much the same as was in the theatre (does not seem to be attenuated down at all). Also, the main channels are not rolled off at all, as still have all the low end information that was in the theatrical mix (alhough it is obivious that most home consumers' center or surrounds reproduce anything below 80hz). So, when playing back the mix in Bass Management, there is a HUGE amount of level driving the subwoofer, since there is all this added low end from the mains along with the high level LFE channel (which i understand also gets boosted +10db). Leaving the mains full bandwidth would make sense with bass management, if the LFE wasn't so hot, but it is.

To make it worse, wouldn't this Bass Management sound terrible if you listened on a subwoofer that was only cal'd as an LFE channel, without considering for 5 channels of low end build up being sent to it as well?
My concern is, that this is causing consumers to turn their OVERALL level down, because the subwoofer is so loud, thus throwing off the overall dynamic listening experience of the movie, dialog may be too quiet and "good" loud parts will lose punch.

Thoughts or Solutions?

If you have dialnorm at -27 when encoding AC3 (which is usually the default), then you are instructing the player to attenuate the signal by 4dB in playback. If this is the case, try with dialnorm -31, then it should produce similar output to PCM (also, see if your player has Dynamic Range Compression, and turn it off - that could also change the level). I suppose it is the same with Blu Ray, but I don't have experience with it yet.....

I have personally not noticed this problem on my small home system (I _expect_ it to distort anyway, so I turn off my distortion-noticing brain circuits.) But in theory it makes sense, that the content from the mains will interfere with the LFE channel. I hope others have some words of wisdom on how to mix and monitor low frequencies (and especially Georgia - she is THE sub authority
I think you should make a new thread out of that question, you'll probably get more answers than here.....