Jump to content

Joel Pinteric

Recommended Posts

Hey all,

I'm curious what mixdown levels most (post-production) sound mixers use with films, particularly short films. I vaguely remember from college that our outputs should be peaking around -10 or -12dBFS at the most. However I think this must vary a lot depending on the application of the film and more confusion is added on what is an acceptable loudness, since peaks do not define loudness. Furthermore I suspect these levels would be inclined differently for web purposes.

I've done a few sound mixes in the past that clients were happy with, following the above rule, however recently I've received a request to basically get the mixes peaking as loud as possible (close to 0dBFS) as my mix was too "quiet". I also went to a student short-film screening at a well known film school and the sound was horribly loud in some films. The films I did sounded acceptable but I hope that there's not a loudness war going on here.

Any background appreciated, even if you have links to post-production mixing/mastering in general.

Thanks a lot.

Link to comment
Share on other sites

Average levels should be at -20dBFS, peaks at -12. Whoever told you to to get levels closer to 0db doesn't know what they are talking about, plain and simple. Music recording pushes the 0db limit but that just isn't the way dialogue is mixed. As you noticed the student films sounded too loud, that's because they are pushing close to 0db and probably compressing it to high heaven.

Link to comment
Share on other sites

Here's the actual thread on gearslutz (in a sticky) that gives you even more information than you want:

http://tinyurl.com/ya6snft

However, the above specs are for post. Levels for location recording depend upon who your tracks will be delivered to and how they will be handled post-wise. If they're for a network, then the -20dBfs (for average levels) and -10dBfs (for peaks) usually applies. If it's for film post, then -20 is a still a good level for program average, and beyond that, keep it fairly consistent (at least, within a given scene) and keep the peaks out of the red. Post will then massage the tracks as need be.

If post doesn't know what they're doing (as is sometimes the case on low budget films), they're capable of mangling your tracks no matter what levels you use, and if they do know what they're doing, they'll be delighted to get clean, consistent tracks.

And, yes, it's not uncommon for amateurs to mistakenly deliver films normalized with peaks up to the red since they've found that music is mixed like that.

Link to comment
Share on other sites

The weird thing about post (especially dailies) is, the editors almost never complain about production tracks that are a little loud; but they do complain about tracks that are too soft. I think peaking mix track dialog at least at -10dBfs is a good idea. The isos can always be a little more conservative.

--Marc W.

Link to comment
Share on other sites

The weird thing about post (especially dailies) is, the editors almost never complain about production tracks that are a little loud; but they do complain about tracks that are too soft. I think peaking mix track dialog at least at -10dBfs is a good idea. The isos can always be a little more conservative.

--Marc W.

+1 I'm a fan of letting the isos be a little lower than the mix track. One "post" guy complained to the director that the location recording was way too quiet because the average level was at -20 and the hottest signal was at -6 (a hand clap). The director evidently looked right at the post guy, thanked him for his time but that his services wouldn't be needed any more and let him know that those levels were exactly where they should be.

Link to comment
Share on other sites

AFAIK, for stereo mix broadcast submissions, -20dBFS ref. level with program peaks not exceeding -10dB still apply. (at least in NYC)

LKFS metering levels are most important for broadcasters, who should adhere to the CALM act. (which has a little to do with acquisition recording...except for the obvious)

There does not seem to be any 'standard' audio level for DVDs and BDs, though I would not suggest peaking it to 0.0dBFS and hammering it peak limiting for sake of louder than then so and so, like in a music CDs. I generally master DVDs to get an RMS level of -25 to -27dB with peaks not exceeding -6.0 dB, though I will certainly get LKFS metering software before the next project.

Link to comment
Share on other sites

AFAIK, for stereo mix broadcast submissions, -20dBFS ref. level with program peaks not exceeding -10dB still apply. (at least in NYC)

LKFS metering levels are most important for broadcasters, who should adhere to the CALM act. (which has a little to do with acquisition recording...except for the obvious)

There does not seem to be any 'standard' audio level for DVDs and BDs, though I would not suggest peaking it to 0.0dBFS and hammering it peak limiting for sake of louder than then so and so, like in a music CDs. I generally master DVDs to get an RMS level of -25 to -27dB with peaks not exceeding -6.0 dB, though I will certainly get LKFS metering software before the next project.

It is very dangerous to assume anything about delivery level specs if you are delivering to a network. You MUST get the spec that the producer agreed to abide by in their contract for the show. I will not start work on a show until I see this spec, in the past a lot of work has had to be done over because assumptions were made.

phil p

Link to comment
Share on other sites

Another rule of thumb is consider the final destination for the project. If this is something for the web, or a little industrial video that's not for broadcast, I'll tend to push the levels a little harder, knowing it's not going to get much of a mix at the end. If it's for broadcast TV or theatrical release, I'm a lot more conservative about where the mix track levels go.

--Marc W.

Link to comment
Share on other sites

+1 I'm a fan of letting the isos be a little lower than the mix track. One "post" guy complained to the director that the location recording was way too quiet because the average level was at -20 and the hottest signal was at -6 (a hand clap). The director evidently looked right at the post guy, thanked him for his time but that his services wouldn't be needed any more and let him know that those levels were exactly where they should be.

With production dialog recording, assuming that the iso tracks are prefader (as they should normally be), and assuming that the mixer and recorder are properly lined up and the gain structure is optimized, the iso tracks will naturally almost always be lower than the postfader mix track. This is because the level of the prefader iso tracks is determined by the input trim, which is set low enough to allow for peaks that will not be reduced by the fader when riding gain for the mix track. The prefader level that goes to the ISO tracks is raised above unity gain with the fader to bring it up to a more ideal level for the postfader mix track. This is also why the fader will naturally find its usual position to be somewhere in the upper fourth of the fader throw, between the unity gain marker (usually labeled "0") and the top (usually labeled "+12"). Therefore, the iso tracks will typically be between odB and 12dB lower than the mix track. (A notable exception is when using the 24db gain option of the Sonosax SX-ST mixer, which makes that iso track up to 24dB lower if the option is fully utilized to preserve a wide dynamic range on the iso tracks.)

Regarding proper level for digitally recording original dialog tracks (such as on the production set), if the hottest signal is -6dBfs, then the tracks are at least 6dB lower than ideal.

Keep in mind that the reference of -20dBfs was never intended to be a target for dialog levels. It is, simply, the recommended level on a 24bit digital recorder's full-scale peak meter for aligning a constant tone that registers "0" on a VU meter. For example, if using a mixer with a VU meter to feed a 24bit digital recorder, the recorder's input should be adjusted so that a tone registering "0" on the mixer's VU meter also registers -20dBfs (20dB below max) on the recorder's peak meter. If you were to do this, an then use the VU meter to keep the levels generally near OVU, you would see that the peaks on the digital recorder's meter were generally well above -20dBFS, but usually below 0dBfs (digital max). Keywords here are "generally" and "usually", which pretty much sums up why the "VU" in "VU meter" could stand for "Virtually Useless" for production dialog tracks (which is why, thankfully, we have stopped seeing VU meters used in field recording equipment). Logically, this also means that -20dBfs is virtually useless for metering while recording original dialog tracks.

So what's the answer? Make sure your mixer and recorder are lined up using a tone (it doesn't matter if the tone is -20dBfs or -30dBfs or -5.5dBfs, as long as the mixer equals the recorder. Then keep the loudest peaks approaching but safely below 0dBfs and use your ears to gauge the desired dynamic range.

Glen Trew

Link to comment
Share on other sites

Hi Mike,

Regarding the +20dBfs comment, I'm sure you are referrencing the Nomad "Never Clip" feature. It is a complex topic and interesting feature that works and has good value, but, for now, to ward off any misconception, suffice it to say that even with Never Clip you cannot record above 0dBfs.

Link to comment
Share on other sites

Here is the approach that I use in post mixing that I find works fairly universally (bear in mind I am old school and used to spend my mornings cleaning headblocks on 1/4 inch 2 tracks and 2 inch 24 track recorders)

I install the Essential Meter Bridge plug in (post compression) on a master fader channel (in ProTools) and calibrate 0VU to equal -18dBFS. I then mix to the VU meters as I would have in the old analogue days maintaining my average levels between -2 to +2 VU based on what my ears are hearing. This pretty much guarantees that my average levels are around -20 and peaks, when they occur should not surpass -10 given I have a gentle limiter sitting in the chain with a threshold set to catch any transients.

Link to comment
Share on other sites

In answer to Glen's and Mike's postings:

Your advice is fully correct if you record with 16bit word length. If you work with 24bit it is a possible workflow, but in my opinion not the best.

1. With 24bit you have a dynamic range of 144dB within the digital domain. Even with a digital headroom of e.g. 24dB, you'd still have 120dB left. That's more than enough to work with. (And what microphone/pre-amp combination has such a SNR - besides digital microphones?)

So there is no technical advantage on getting my peaks as close as possible to 0dBFS.

(With 16bit you only have 96dB dynamic range. In this case you have to record as hot as possible to get a good SNR.)

2. Besides my technical argument there is another reason I recommend not to get your peaks up to 0dBFS. - It's clipping protection.

As we all know digital clipping is not as nice as analog clipping. Once a peak hits 0dBFS, your signal is bad. If you keep your peaks right under 0dBFS you constantly have to re-adjust your gain. Keeping a digital headroom makes your life easier without compromising the sound quality.

3. 0VU is not just there for aligning a constant tone. A VU meter is a meter that equals our hearing perception better than a peak meter. We don't hear peaks. We hear "VU". Of course VU is not the best system therefore. But it is cheap in analog equipment and easy to build in the digital domain. BTW: Have a look at the upcoming LU (Loudness Units).

So if we keep our levels at 0VU (e.g. -20dBFS) we provide signals at a constant loudness. (Depending on the post-pro specifications we limit/compress our signal.) If you set your levels keeping the peaks right under 0dBFS, depending an the signal your loudness may vary. - More work in post!

There are more arguments not to go all the way up to 0dBFS, but in between I have to do something for my living. :-)

Link to comment
Share on other sites

I install the Essential Meter Bridge plug in (post compression) on a master fader channel (in ProTools) and calibrate 0VU to equal -18dBFS.

1990s software VU meters? Yikes!

Try the Dorrough meters -- they'll give you more of a real-world idea of where average loudness is at. A much more accurate (albeit much more expensive) choice is the Dolby Media Meter, a plug-in that will provide ITU-R (LKFS/LUFS) levels. A lot of distributors now require this for final mixes, especially the Discovery Channel and HBO. This is what I see being used for a lot of mixes for broadcast delivery these days.

For much less money, the Audioplugger K-Meter is pretty good and is only $50. It takes a little while to figure it out (showing both average and peak levels simultaneously), but they claim to be able to handle the ITU spec.

kmeter-1-4-0-big.jpg

--Marc W.

Link to comment
Share on other sites

" a gentle limiter "

??

" (And what microphone/pre-amp combination has such a SNR - besides digital microphones?) "

Hi, Tom, and welcome...

the quote above is the operative issue, I understand the technical issues you refer to, but they typically do not apply to production sound for movies and TV...

also, in post production there are usually delivery spec's that must be met, and now we are also getting into loudness regulation for broadcasters.

and of course you also mention differences between VU and peak metering which are also quiet important to understand,...

and viva Dorrough !! (or a good old O'scope!)

Link to comment
Share on other sites

When I do post prod mixing for documentary or TV program (for my local TV station lost in the middle of the pacific), I have a maximum output at -6dBFS, with the average level for voice around -10dBFS. So it always confortable to have the recorded audio at an average of -12dBFS.

-18 or 20dBFS is way too low i think, especially when after the shoot the director check the takes at home listening throught little crappy speakers that need to be crancked up to hear the programs recorded at an average of -20dBFS. Not all listening system are capable of translating audio recorded at -18/20 dBFS without adding noise...

Link to comment
Share on other sites

In answer to Glen's and Mike's postings:

Your advice is fully correct if you record with 16bit word length. If you work with 24bit it is a possible workflow, but in my opinion not the best.

My earlier post about 0VU reference and it's reason for being was not meant to be advice or an artistic recommendation, but a technical statement of fact, which I'll try again with a little more clarity: With regards to recording original production tracks, the only reason for a 0VU reference is to allow the alignment of equipment used in that recording (during production or post production) that may have VU meters. It is not, and was never intended to be a target for recording levels. This is no more true with 24bits than 16bits or 12bits.

2. Besides my technical argument there is another reason I recommend not to get your peaks up to 0dBFS. - It's clipping protection. As we all know digital clipping is not as nice as analog clipping. Once a peak hits 0dBFS, your signal is bad.

The ONLY reason to keep the peaks below max is for clipping protection. The term "clipping" is borrowed from the solid-state analog phenomenon, and not quite what happens when digital max is exceeded, but is certainly no more noticeable or more offensive than the clipping of an analog audio signal. I wonder if, in your post, there might be confusion between "analog clipping" and the less objectionable distortion that occurs when overdriving analog tape (which does not clip) or vacuum tubes.

3. 0VU is not just there for aligning a constant tone.

The statement about aligning was not about 0VU, but “0VU reference” for a full-scale peak meter used on digital equipment (generally accepted these days as -20dBfs for original tracks). "Constant tone” was specified because very short tones do not fully register on a VU meter. Try recording a xylophone while monitoring with VU meter showing normal deflection, and you'll get almost constant overload distortion, which is another illustratation of why "VU" stands for "Virtually Useless" for original production tracks). Yes, for the purpose of recording original production tracks, the only reason for the existence of a 0VU reference is for aligning digital meters with VU meters (which is rarely done anymore with the tone we put on our original production tracks). If it weren't for that need, there wouldn't have to be any markings or scale on the dBfs peak meter at all. You'd only need to avoid max and use your ears to judge how much to ride gain or compress to achieve the desired dynamic range. Of course, the motion of the meter display gives you an idea of how close you are getting to max level and how much margin of safety you might need to avoid it, but the hash marks below 0 are of no use for this. I’ll add an obvious point and say that an alignment tone is needed for aligning one piece of analog equipment to another, and an analog recording to the input of an analog device, but this tone could be any level, as long as it is specified what the level is.

Speaking of using your ears, in the analog Nagra days (which had a 0VU reference of 12dB below it’s max level (-8 on the modulometer plus 4dB above the modulometer’s “0”)), production mixers generally had their headphones turned down a bit lower than they do now. This is because the signal level was generally louder due to the reduced dynamic range that a 0VU reference of 12dB below max allow. I’ve heard many sound mixers say that the headphone volume knob on the same mixer they used in the Nagra days is now turned up a bit higher. The comment is usually used to suggest a loss in hearing sensitivity from years of using headphones. Though this may be partially true, even someone without a change in their hearing sensitivity would turn up their headphones when changing from a 0VU reference of -12 to -20.

This is a very complex technical issue that requires more space than should be allowed on this forum (it could – and should – be a book), so it is impossible to fully explain here so that it makes sense to those with varying degrees of understanding on the topic. I stand by all the statements above, but, as always, welcome the chance to be shown I am wrong.

Glen Trew

Link to comment
Share on other sites

Thank you, Glen.

This topic gets rehashed here every once in a while, and always makes me question my choice to record "hot". It's nice to have your voice confirm that my choice is just fine. And I don't recall ever getting a call that said my levels were too high.

I am very critical of the limiters on one bit of gear, however, so I do get upset when I get "caught" on my mix. But I don't feel like dropping my levels for those once-in-a-while moments. At least they've got the ISO at a lower level.

Robert

Link to comment
Share on other sites

Anyone use TC Electronic's "Radar Meter"? They have it available in a touchscreen monitor format

http://www.tcelectronic.com/touchmonitor.asp

or Pro Tools HD plugins, but I am waiting for the release of the forthcoming LM6 for native DAWs, as it supports LKFS metering.

http://www.tcelectronic.com/lm6.asp

Also for those of you who use software meters, which ones have a GUI that can be expanded to fill a whole screen? What I'd like to do is to place a smallish monitor, like a 15" to 22" widescreen or something, underneath my projection screen, so that it always has a full screen meter graphic on it. The reasons that I'm looking at this meter, is that I like the way it looks, presents information, and is one of the plugs that is going to support the new AAX format from Avid for that time that I move to Pro Tools 10 HD.

Link to comment
Share on other sites

I've done a few sound mixes in the past that clients were happy with, following the above rule, however recently I've received a request to basically get the mixes peaking as loud as possible (close to 0dBFS) as my mix was too "quiet". I also went to a student short-film screening at a well known film school and the sound was horribly loud in some films. The films I did sounded acceptable but I hope that there's not a loudness war going on here.

The answer to "how loud should I mix?" is dependent on the subject matter and the destination. My experience in post production has been almost entirely feature length films that receive movie theater mixes, so I can only comment about those kinds of mixes.

Talking about maximum db levels is the wrong approach. The only limit in our digital audio world is 0dbfs. Instead we should be talking about average listening levels. What audio level do you find most appropriate to listen to movies like "The King's Speech" or "The Social Network"? Now how about "Sucker Punch" or "Batman: The Dark Knight"? You may be tempted to turn up the volume on the latter movies, but maybe not for the former. In a dedicated home theater environment I find myself setting the volume around the same setting for most movies, making only 1-2 db adjustments. How can this be when these movies have drastically different mixes? Movie mixes maintain a high headroom to allow for loud passages. Quieter dramas have this headroom even if they don't use it. There is an established average loudness that most movie mixes conform to.

How do you figure out what this average loudness is and how do I setup my system? There is an excellent book (http://amzn.com/1458400395) that walks you through how to do this. I highly recommend buying it if you are interested in configuring a surround sound system for editing and mixing movies. The process involves calibrating your D to A device (Digidesign/Avid 192 I/O in my case), using a volume controller, and determining the proper gain structure and listening levels. In my setup I have a Dangerous Music Surround Monitor Controller System (http://www.dangerous...c.com/stsr.html) that ensures a highly accurate and repeatable reference level setting. I set the volume knob to the same setting every time I'm editing/mixing and I don't change it.

If you are mixing for TV, video, web or elsewise, I'd say there is no standard level for mixing. A TV series like "Lost" is mixed different than "Wheel of Fortune" which is mixed different than "Mythbusters". "Lost" is mixed almost theatrically (see a video here: http://youtu.be/ct8gNleLemE). The other two almost undoubtedly louder, although television networks and channels still apply audible compression, which is a whole different issue.

So much can be said - it's really a can of worms. I'm happy knowing that the effort I put into educating myself about mix levels and the time spent properly setting up my system has produced mixes that translate very well in both commercial movie theaters and home viewing environments.

Mark O.

Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

×
×
  • Create New...