Dolby Digital/AC3 Metadata

[Enos]

Hi guys! I've been scouting different threads around these forums trying to find a specific answer to my question but since i don't seem to find one I will open a new thread.

My question is regarding the process of encoding a mix (be it stereo or surround) as an AC3 file for a Dolby Digital print. I understand the basic workings of this process but have several specific questions regarding the options available in the process regarding dialnorm and dynamic range control.

In the dubbing suite I mix in, the room is calibrated for theatrical mixing standards on which a pink noise signal at -20dBFS = 85dBSPL at the mixing position. When mixing, my average dialogue levels rest around -20dBFS with lower bits going an average of 6dB below and the louder bits of dialogue peaking anywhere between -15 up to -10 or even a bit higher at times.

Anyways, moving on, when I finish my mix and have it all bounced down to 6 discrete files I move on to encoding using Apple's Compressor software. I import the discrete channel files into the program and my question regards the settings that follow.

Please correct me if I am wrong in any way:

Regarding dialnorm, if I my average dialogue (RMS) lies at say -20dBFS, then I should set the dialnorm value to -20 which would mean my mix would be reduced by 11dB once dialnorm is applied right?

So, if I cheat this and actually set the dialnorm value to say -31, would the encoding effectively blindly take in my settings and result in a louder mix? Or is there something that will still go wrong. basically, is it possible to cheat the dialnorm reduction?

Also, regarding dynamic range control, when encoding for DVD is it common to apply a light profile of DRC such as Film Light or Standard? I guess if you are printing for theatrical release you would set DRC to None but for DVD would you possibly be applying a little bit of it considering it will probably be played back in homes (louder environments than a cinema).

Finally, the film i am currently mixing is due for online release. However, I have been treating it as if it were for theatrical release using -20dBFS as my dialogue average level. I know I probably have done wrong to mix it like this and should have mixed hotter from the start. So now I will have to tighten my dynamic range and have a few questions:

Would it be wise to automate my master fader within protools prior to bouncing and thus raise the quiter bits closer to the louder ones as this sounds to me as a slightly more natural approach than applying a compressor/limiter to the outputs? What level would I then be aiming for and where do you recommend my dialogue average should lay?

One final thing....and thansk in advance for any possible replies I might get!

Once I have bounced this web version of the mix and am ready to encode how would the encoeding parameters be affected due to the nature of a web release and the format of playback it will most likely be heard on (laptops and desktop speakers ...I know :( )

I still haven't got the brief regarding what final format will be delivered to be uploaded online and am wondering what that might be? Being a stereo mix, it wouldn't be a Dolby Digital 2.0 encoding right? If it is...then how would dialnorm and DRC be set for such a scenario? Should I not bother tightening the dynamic range in the mix and just leave Dolby's DRC do it and set a more heavy DRC parameter in the encoding stage or should I take care before the bounce and try to avoid DRC altogether?


Ok...enough questions...Any help would be greatly appreciated!!!


Enos

[danijel]

Originally Posted by Enos

In the dubbing suite I mix in, the room is calibrated for theatrical mixing standards on which a pink noise signal at -20dBFS = 85dBSPL at the mixing position. When mixing, my average dialogue levels rest around -20dBFS with lower bits going an average of 6dB below and the louder bits of dialogue peaking anywhere between -15 up to -10 or even a bit higher at times.

If your dialogue RMS is -20dBFS, I suspect something is wrong with the calibration. Average dialogue should be more towards -27dBFS RMS. Of course, this is just a shot in the dark, your movie might be different, with a lot of shouting and VO-ish close-up statements.....?

Originally Posted by Enos

Regarding dialnorm, if I my average dialogue (RMS) lies at say -20dBFS, then I should set the dialnorm value to -20 which would mean my mix would be reduced by 11dB once dialnorm is applied right?

So, if I cheat this and actually set the dialnorm value to say -31, would the encoding effectively blindly take in my settings and result in a louder mix? Or is there something that will still go wrong. basically, is it possible to cheat the dialnorm reduction?

You can't cheat, because DRC threshold in players is also set in regards to dialnorm. So your sound could be louder overall, but it would get compressed as hell. (ie dialogue will be turned down all the time by DRC, with probably only quiet ambiances swimming out from time to time.)

Originally Posted by Enos

Also, regarding dynamic range control, when encoding for DVD is it common to apply a light profile of DRC such as Film Light or Standard? I guess if you are printing for theatrical release you would set DRC to None but for DVD would you possibly be applying a little bit of it considering it will probably be played back in homes (louder environments than a cinema).

You're not applying compression at encoding stage, you're merely telling the player what kind of program lies on the audio track, so it knows how to handle the sound depending on the playback environment. Whatever you set DRC to (including 'None'), most home and software players will compress the hell out of your mix anyway.
Oh, by the way, you do not use the same encoder for theatrical release. Dolby Digital track on film print also utilises AC3 technology, but the tracks are not compatible with DVD AC3. It's made with Dolby DMU (Digital Mastering Unit).

Originally Posted by Enos

Finally, the film i am currently mixing is due for online release. However, I have been treating it as if it were for theatrical release using -20dBFS as my dialogue average level. I know I probably have done wrong to mix it like this and should have mixed hotter from the start. So now I will have to tighten my dynamic range and have a few questions:

Your questions regarding dynamic range for online release depend on what 'online release' means. Will it be streamed in a browser window (youtube?), or will it be made available for download in higher quality? There are no standards in this regard, and oppinions differ, but I would personally squeeze the DR for streaming, and leave it as is for HQ download.

Originally Posted by Enos

I still haven't got the brief regarding what final format will be delivered to be uploaded online and am wondering what that might be? Being a stereo mix, it wouldn't be a Dolby Digital 2.0 encoding right? If it is...then how would dialnorm and DRC be set for such a scenario?

It will probably be H264 with AAC or MP3 audio, so no DRC, dialnorm or Dolby.

[Enos]

If your dialogue RMS is -20dBFS, I suspect something is wrong with the calibration. Average dialogue should be more towards -27dBFS RMS. Of course, this is just a shot in the dark, your movie might be different, with a lot of shouting and VO-ish close-up statements.....?

Hmmm...I'm quite new to mixing and working in a calibrated room,etc... For some reason I understood -20 to be normal average dialogue level. Sp -27 is a more usual RMS level for dialogue then?

You can't cheat, because DRC threshold in players is also set in regards to dialnorm. So your sound could be louder overall, but it would get compressed as hell. (ie dialogue will be turned down all the time by DRC, with probably only quiet ambiances swimming out from time to time.)

Hmmm..but can't you deactivate DRC when encoding? Or will the player apply DRC of its own?? What I mean is, if I deactivated DRC upon encoding and set the dialnorm value to -31 (even though it really was higher). Then unless the player applies values of its own as opposed to those set into the metadata of the AC3, DRC should not be applied as the set dialnorm is under threshold level?

You're not applying compression at encoding stage, you're merely telling the player what kind of program lies on the audio track, so it knows how to handle the sound depending on the playback environment. Whatever you set DRC to (including 'None'), most home and software players will compress the hell out of your mix anyway.
Oh, by the way, you do not use the same encoder for theatrical release. Dolby Digital track on film print also utilises AC3 technology, but the tracks are not compatible with DVD AC3. It's made with Dolby DMU (Digital Mastering Unit).

Why would they compress my mix? Don't they simply stick to Dolby standards. You mean home DVD players each do their own compressing and manipulation of the audio??

Oh and thanks for the info regarding DMU So DMU is what you get when you are paying the license and getting a tech engineer in?

Your questions regarding dynamic range for online release depend on what 'online release' means. Will it be streamed in a browser window (youtube?), or will it be made available for download in higher quality? There are no standards in this regard, and oppinions differ, but I would personally squeeze the DR for streaming, and leave it as is for HQ download.

By online release I mean live streaming in a browser window via Dailymotion. They are allocating a larger part of their server than usually allowed to users for our release as we are giving them the exclusivity for the first few weeks.

Last time we released it via Dailymotion, we had Youtube knocking on our door asking for permission to take the film on to their servers too, as we had about 3mill views in the first month and half! So they hosted the film on their servers allowing for the entire 40 minutes of film to run at once at a decent quality. Anyways, this time its a full feature being released and no downloads available due to copyright reasons. So by squeezint the DR, what kind of DR would you be talking about? More or less?

It will probably be H264 with AAC or MP3 audio, so no DRC, dialnorm or Dolby.

Cheers! I'll ask about it today ..just to make sure!!

[georgia]

I just added some info about this to my thread: It might help you.

Geo's sound post corner

last couple of posts...

cheers
geo

[Enos]

Thanks Georgia! I'll run and check it out straight away

[danijel]

Originally Posted by Enos

Hmmm...I'm quite new to mixing and working in a calibrated room,etc... For some reason I understood -20 to be normal average dialogue level. Sp -27 is a more usual RMS level for dialogue then?

Here's a quote from Tom Holman:
Almost all Hollywood movies are made such that the average level of A weighted dialogue will be -27 dB re Full Scale.

Go through the stickies at the DUC and here again a couple of times, sleep it over, and it will all get sorted out

Originally Posted by Enos

Hmmm..but can't you deactivate DRC when encoding? Or will the player apply DRC of its own??

First, in AC3 metadata, there is no way to deactivate DRC - you only indicate the type of program on the dics. If you set it to 'None', it means you did not tell the player about what is on the dics, and it will apply its default DRC setting - it does not mean that 'None' compression is to be applied.
The only way to deactivate DRC is in the player's menu, and even that is not possible with all players.

Originally Posted by Enos

What I mean is, if I deactivated DRC upon encoding and set the dialnorm value to -31 (even though it really was higher). Then unless the player applies values of its own as opposed to those set into the metadata of the AC3, DRC should not be applied as the set dialnorm is under threshold level?

Dialnorm DETERMINES the threshold in the player, so you set it to -31, and your dialogue is at -20, the dialogue will trigger the DRC all the time.

Originally Posted by Enos

Oh and thanks for the info regarding DMU So DMU is what you get when you are paying the license and getting a tech engineer in?

Yes.

Originally Posted by Enos

By online release I mean live streaming in a browser window via Dailymotion.
..........
So by squeezint the DR, what kind of DR would you be talking about? More or less?

From my tests, I've concluded that computers with poor speakers (including laptops), don't have sufficient power to deliver dialogue at decent listening level if the mix is dialnorm -27. I'd go for at least -23, or better yet, leave it at -20 if that's where you are.
Regarding peaks, do a search for 'youtube audio pumping' or something like that. AFAIR, peaks should be kept under -3dBFS or so because of their encoding 'bug', so as your mix will eventually end up there, it is an important consideration.

Originally Posted by Enos

They are allocating a larger part of their server than usually allowed to users for our release as we are giving them the exclusivity for the first few weeks.

Last time we released it via Dailymotion, we had Youtube knocking on our door asking for permission to take the film on to their servers too, as we had about 3mill views in the first month and half! So they hosted the film on their servers allowing for the entire 40 minutes of film to run at once at a decent quality. Anyways, this time its a full feature being released and no downloads available due to copyright reasons.

You are talking about the independent LOTR thing, right? Looks like good energy went into that project! thumbsup

[danijel]

Oh, and forget Dolby, AC3, dialnorm, DRC etc for this release. Try to talk to someone at Dailymation, or better yet, do your own tests - upload a clip with sines (at different levels), dialogue etc, and see how it turns out in the playback (you can perhaps record it back into your DAW). As I said, youtube has some DR compression algorithm, but Vimeo, for instance, leaves the DR untouched.

[Enos]

Here's a quote from Tom Holman:
Almost all Hollywood movies are made such that the average level of A weighted dialogue will be -27 dB re Full Scale.

Go through the stickies at the DUC and here again a couple of times, sleep it over, and it will all get sorted out

Indeed, I had it wrong! Thanks for the info! I've been lucky this time though as -20 dialnorm is probably good considering its for online release and will be played back on laptops quite a lot :p

First, in AC3 metadata, there is no way to deactivate DRC - you only indicate the type of program on the dics. If you set it to 'None', it means you did not tell the player about what is on the dics, and it will apply its default DRC setting - it does not mean that 'None' compression is to be applied.
The only way to deactivate DRC is in the player's menu, and even that is not possible with all players.

Again, I had got that totally wrong! I thought turning the DRC profile to None was like deactivating DRC and telling the player to not apply any! Wow...important thing to get right then!! :s

Dialnorm DETERMINES the threshold in the player, so you set it to -31, and your dialogue is at -20, the dialogue will trigger the DRC all the time.

Yea...it all makes sense now, knowing that DRC can not be turned off in encoding. It also makes perfect sense that cheating dialnorm and DRC is not possible either

From my tests, I've concluded that computers with poor speakers (including laptops), don't have sufficient power to deliver dialogue at decent listening level if the mix is dialnorm -27. I'd go for at least -23, or better yet, leave it at -20 if that's where you are.
Regarding peaks, do a search for 'youtube audio pumping' or something like that. AFAIR, peaks should be kept under -3dBFS or so because of their encoding 'bug', so as your mix will eventually end up there, it is an important consideration.

Great! I've been roughly mixing with dialnorm at -22 to -20 with peaks slightly higher than that! I definitely have to look into what youtube or dailymotion does to the mix before I print the master! I recently did another LOTR fan film called the Hunt for Gollum but I didn't do the final master. I sent away the final mix in dialogue,fx & music stems. Anyways, when released on dailymotion it was VERY quiet and slightly compressed too. It didn't sound anywhere near to what it did in the mixing suite. This film though is slightly different and apart from having gained experience, I will be supervising and present in the final printing and rendering of everything so hopefully it will be fine!

You are talking about the independent LOTR thing, right? Looks like good energy went into that project!

Yes..well i did the mixing for another LOTR fan film earlier this year called The Hunt for Gollum. It did really well. This film I have been working on for the alst few months is a new LOTR fan film called Born of Hope (Born of Hope). For this film I have been supervising the sound department, editing sound effects and backgrounds plus doing the final mix. We've had about 7 more people in the sound team including production mixer, dialogue editor and foley sup + artists.

Anyways, two more day to deadline! Tomorrow is final tweaks and test bounce to then playback on several different laptops and playback systems. Wednesday is final printmaster and Wednesday night we render picture and sound for the final print! Thursday we deliver the film and then December 1st is the release!! Then I can get some sleep!! :p


Anyways, thanks a LOT for the info...I thin I understand the basics of dolby encoding...at least regarding Dialnorm and DRC

[kk@jamsync.com]

Originally Posted by danijel

First, in AC3 metadata, there is no way to deactivate DRC - you only indicate the type of program on the dics. If you set it to 'None', it means you did not tell the player about what is on the dics, and it will apply its default DRC setting - it does not mean that 'None' compression is to be applied.
The only way to deactivate DRC is in the player's menu, and even that is not possible with all players.

That ought to be printed out, framed, and tacked to the wall in any studio that does encoding. Really!

[Enos]

Indeed! This stuff is SO important and it is SO easy to really screw up hours of work if this is not done correctly!

Thanks again for all the information!

[danijel]

Originally Posted by Enos

Anyways, two more day to deadline! Tomorrow is final tweaks and test bounce to then playback on several different laptops and playback systems. Wednesday is final printmaster and Wednesday night we render picture and sound for the final print! Thursday we deliver the film and then December 1st is the release!! Then I can get some sleep!! :p

Anyways, thanks a LOT for the info...I thin I understand the basics of dolby encoding...at least regarding Dialnorm and DRC

Good luck, it's a pleasure to give a little help on such a great project thumbsup

[Enos]

Thanks indeed!

By the way, I have been figuring out what formats we must deliver for the online release. As the premiere release will be on Dailymotion they will take a stereo AAC file at 256kbit. At least that is better than if we were putting up to youtube!

I'm looking into the ins and outs of the AAC format and preferences available to it. Anyone know anything that might be of importance regarding what it does or does not do to audio?