Mixing levels for Dolby Digital: DVD vs. theatrical

[Emre Safak]

Is the recommended mixing environment of -20dBFS RMS pink noise at 85 dbC valid for DVDs as well as film? I calibrate my system to this spec (with the surrounds 3dB lower) yet my DVDs sound quieter than commercial products. Is it because I edit in a cramped environment? Do they use more aggressive compression ?

[Steve House]

Originally Posted by Emre Safak

Is the recommended mixing environment of -20dBFS RMS pink noise at 85 dbC valid for DVDs as well as film? I calibrate my system to this spec (with the surrounds 3dB lower) yet my DVDs sound quieter than commercial products. Is it because I edit in a cramped environment? Do they use more aggressive compression ?

Are you setting your meters to read -20dBFS on the pink noise signal? If so that could be the problem. According to my info, the proper procedure is to do each channel one at a time and set the DAW levels so each channels output meters read -20dBFS when fed a -20dBFS reference sine-wave tone. Switch to -20dBFS RMS pink noise without changing the level settings. Peak reading meters will read about -10dBFS at that point but don't reduce the levels. Adjust each monitor in turn, muting all the others and set the monitor amplifier levels to produce an SPL of 81dB on each channel, setting all speakers to the same level if going to DVD release. If going to broadcast or analog videotape, use 79dB SPL instead. For theatrical release use 81dB SPL for the fronts and centre and reduce the surrounds by 3dB to 78 dB SPL. (Source: Holman, "Sound for Digital Video," pp 271-273)

[Emre Safak]

-20dbFS RMS (and about -10dbFS peak) is a property of the pink noise signal. I use a ready-made file for that. Then I hard pan it to every channel in a surround project using Vegas.

Finally I adjust the volume of the speakers to read 85dbC on the sound level meter.

[Steve House]

Originally Posted by Emre Safak

-20dbFS RMS (and about -10dbFS peak) is a property of the pink noise signal. I use a ready-made file for that. Then I hard pan it to every channel in a surround project using Vegas.

Finally I adjust the volume of the speakers to read 85dbC on the sound level meter.

You might be setting the monitor volume a bit high. This would lead to you to turn down overall levels when mixing by ear as it sounds too loud. Your level is about 4-6dB hotter than recommended and a difference of 6dB is twice the loudness. Holman reccommends ~80dBSPL as noted above. Also be sure to adjust one channel at a time with all the others muted.

[Tomlinson Holman]

Gentlemen:

I am impressed by the correct information given. However, there is one other factor. Dialnorm, built into AC-3 by the requirements of ATSC normalizes dialogue. Almost all Hollywood movies are made such that the average level of A weighted dialogue will be -27 dB re Full Scale. Dialnorm pushes this down by 4 dB, "normalizing" all dialogue to -31 dBFSrms. So small rooms calibrated to 85 dB SPL and playing DVDs from a Dolby decoder (does not apply to DTS) will play at 81 dB equivalent.

When you find your levels are lower than this, I'm afraid I don't understand why if you have calibrated as suggested. My students all mix dialogue within ±2 dB of each other on the Spielberg Scoring Stage in the control room at USC, which is calibrated to 81 dB. We find this interchanges with a much larger theater calibrated at 85 because everyone who has done it has found that the same SPL in a larger room sounds less loud than in a small one.

Hope this helps.

[Emre Safak]

Thank you for weighing in, Mr. Holman.

[Emre Safak]

I found the problem: I was mixing by ear without calculating the RMS level. I discovered that they were rather lower than they should be, even though they sounded right. Perhaps this is because, as Steve suggested, I have set my monitors' volume too high. I think 85dbC SPL is just too much for my cramped room. Perhaps if I set it 5-10dB lower it would sound right and measure right?

Meeting the -27dBFS dialog spec with a recording on location is pretty difficult if you want to use any compression unless the track is very clean, otherwise the compression is going to pull the noise floor up to audible levels. My procedure is to run a round of noise reduction, normalize to -3dBFS, then adjust the graphic dynamics to pull the noise down while adjusting the rest to meet the -27dBFS RMS level. Is this a good way of going about things?

[Lee Sharp]

Hi guys

Just to clear up my own confusion

Line up and peak

for NTSC -20dbfs - -10dbfs
PAL -18dbfs - -10dbfs

Before any post production consideration, is there any level that dialogue should not drop below during the prodcution stage?

For example say a scene has a peak at which point an actor shouts, the shout would then be around -12/-10dbfs, but this will make the rest of recorded dialogue somewhere closer to -20dbfs... so is there any level where it should NOT drop below, from a production mixer perspective?

Also ive seen conflicitng line ups and peak level

Majority say EBU line up is -18dbfs SMPTE -20dbfs, but ive seen another document that sepcifies SMPTE = -24dbfs, which makes it -20Dbfs at 0VU not 0dbu

If anyone can clear these areas up I would really appreciate it


Sharpy

[Rick Reineke]

For the major networks here in NY land. Normally, the guidelines are: 1KHz. @ -20dBFS , with program audio peaks not exceeding -10dBFS. I've pushed RMS levels beyond rationalization, advertising clients pressure' "we want it in your face... FU__'ING LOUD".
With all the talk about standardizing loudness levels... A sad state of affairs, for sure. ($$$$ talks)

[Lee Sharp]

Originally Posted by Rick Reineke

For the major networks here in NY land. Normally, the guidelines are: 1KHz. @ -20dBFS , with program audio peaks not exceeding -10dBFS. I've pushed RMS levels beyond rationalization, advertising clients pressure' "we want it in your face... FU__'ING LOUD".
With all the talk about standardizing loudness levels... A sad state of affairs, for sure. ($$$$ talks)

But if your gain struture is set so th loudest nosie peaks at -10dbfs, where then is the majority NOT supposed to go beyond.

Lets use the screaming example, Theres conversation and then a scream, the gain is set so the scream peaks around -10dbfs, but then the average dialogue level could fall anywhere. Is there any level it shouldnt go below?

Everyone always talks about peaks and reference tone, but never where it CANNOT drop below.

[Steve House]

Originally Posted by Lee Sharp

...
Also ive seen conflicitng line ups and peak level

Majority say EBU line up is -18dbfs SMPTE -20dbfs, but ive seen another document that sepcifies SMPTE = -24dbfs, which makes it -20Dbfs at 0VU not 0dbu

If anyone can clear these areas up I would really appreciate it


Sharpy

The lineup point is basically related to the amount of headroom the system will have.

SMPTE: +4dBu = 0VU = -20dBFS, thus 0dBu = -4VU = -24dBFS
EBU: 0dBu = 0VU = -18dBFS

A number of commonly used mixers, including Mackie and Sound Devices, also follow the EBU convention rather than SMPTE in their meter calibration. Since their 0VU output voltage is 4dB less than SMPTE standard 0VU voltage, if you wanted everything to be SMPTE norm you'd line up their tone to -24dBFS and use the +4 VU mark on their meters as your normal recording level. (Coincidentaly, -24dBFS is also the "dialnorm" value that dialog should average in a broadcast mix.) In reality, it doesn't matter until you're outputting to the final delivery stage, as long as you set a standard and adhere to it. The bottom line reason for aligning a mixer to a camera or recorder is so you can look at the mixer's meters and know what is going on in the recorder and your normal recording level doesn't drive you into clipping on the peaks. You also want to be consistent shot to shot and day to day so that a scene whose dialog metered 0VU on Friday will be the same level as a scene shot metering 0VU on Monday.

Originally Posted by Rick Reineke

For the major networks here in NY land. Normally, the guidelines are: 1KHz. @ -20dBFS , with program audio peaks not exceeding -10dBFS. I've pushed RMS levels beyond rationalization, advertising clients pressure' "we want it in your face... FU__'ING LOUD".
With all the talk about standardizing loudness levels... A sad state of affairs, for sure. ($$$$ talks)

Tell your client all that attitude serves is sending the viewer's finger mashing down on the mute button as fast as it can get there, guaranteeing none of his expensive message is going to be heard at all.

[Lee Sharp]

Originally Posted by Steve House

The lineup point is basically related to the amount of headroom the system will have.

SMPTE: +4dBu = 0VU = -20dBFS, thus 0dBu = -4VU = -24dBFS
EBU: 0dBu = 0VU = -18dBFS

A number of commonly used mixers, including Mackie and Sound Devices, also follow the EBU convention rather than SMPTE in their meter calibration. Since their 0VU output voltage is 4dB less than SMPTE standard 0VU voltage, if you wanted everything to be SMPTE norm you'd line up their tone to -24dBFS and use the +4 VU mark on their meters as your normal recording level. (Coincidentaly, -24dBFS is also the "dialnorm" value that dialog should average in a broadcast mix.) In reality, it doesn't matter until you're outputting to the final delivery stage, as long as you set a standard and adhere to it. The bottom line reason for aligning a mixer to a camera or recorder is so you can look at the mixer's meters and know what is going on in the recorder and your normal recording level doesn't drive you into clipping on the peaks. You also want to be consistent shot to shot and day to day so that a scene whose dialog metered 0VU on Friday will be the same level as a scene shot metering 0VU on Monday.


Tell your client all that attitude serves is sending the viewer's finger mashing down on the mute button as fast as it can get there, guaranteeing none of his expensive message is going to be heard at all.

Thanks there Steve thats a lot os usful information :) So dialgoue shouldnt really drop below -24dbfs.

My confusion came from 2 charts

1 was accurate and sed

ebu

0db= -4vu = -18dbfs

smpte

+4db= 0vu= -20dbfs


I got confused when another chart said

SMPTE

0dbu=-4vu = -20dbfs this is clearly the wrong one as you state as the output voltage is not +4dbu

What does that mean in terms of volume between EBU and SMPTE, does Smpte put more higher voltage into a signal to then turn the level down, sort of more for less, where as EBU seems to be less for more...

[Lee Sharp]

Also I forgot to mention

When lining up, Are we using the 0VU to recognise it or dbu? That another spanner in the works or is that upto the individual?

Sound devices mixers appears to be metered as dbu as the meter stops at +20/0dbfs. I know it can be set to read both vu, peka and dbu, but the meter itself would look as if its dbu, and is calbirated to 0vu=0dbu.

but SQN of course use PPM. which also has -4 VU

[Steve House]

Well, 0dBu is an absolute mesasurement while the others are ratios.

0dBu SMPTE = 0dBu EBU = 0.775volts RMS
0VU EBU = 0.775 volts RMS
0VU SMPTE = 1.228 volts RMS

Loudness is a subjective perception and doesn't equate directly to any particular voltage. My speakers may be more efficient than yours or vice versa so sending a +4dBu signal to them might sound quite different from sending the same signal to yours. That's why the final step of setting up a monitor system is to send a standard signal-level tone to the speakers and adjusting their volume controls to a standard sound level as measured on a sound pressure meter at the mix position.

[Lee Sharp]

Originally Posted by Steve House

Well, 0dBu is an absolute mesasurement while the others are ratios.

0dBu SMPTE = 0dBu EBU = 0.775volts RMS
0VU EBU = 0.775 volts RMS
0VU SMPTE = 1.228 volts RMS

Loudness is a subjective perception and doesn't equate directly to any particular voltage. My speakers may be more efficient than yours or vice versa so sending a +4dBu signal to them might sound quite different from sending the same signal to yours. That's why the final step of setting up a monitor system is to send a standard signal-level tone to the speakers and adjusting their volume controls to a standard sound level as measured on a sound pressure meter at the mix position.

Thats great thanks, I was going to mention confusion over EBU 0dbu = -4vu but i just found a text thats cleared that up

That basically EBU = 0dbu = -4vu/0vu. In relilty it is infact -4vu, but the EBU standardise it as 0vu like sound devices to make it more sensible when calibrating, or rather than saying -4vu its still 0vu but 4db quieter.

This makes sense, adn as you say volume is relative, so i guess its no biggy to standardise the numbers down by 4db :)