adjusting audio volume?

[Stanley Szpala]

I need to adjust audio volume in my film, which is meant as a festival submission to be ultimately/hopefully shown in a movie theater. I would appreciate some tips. I'd like to do it nicely - there is nothing worse than watching a film with audio levels overblown as in TV commercials.

I use Sony Vegas Movie Studio Platinum HD 10, and Sound Forge or Audacity for audio editing.

Setting the audio just below 0dB will create unnaturally loud soundtrack ( I think).
The scale for audio volume in the audio monitor in my Vegas is from 0 down to -60. What should be a volume for a conversation? But this is the max volume, not the average...

I thought I should set the levels comparable to the volume of a conversation in a film from a dvd, but I'm not sure what workflow would work. I can't adjust the volume through playing that dvd in PowerDVD or a similar player, as I don't know how to correlate the volume levels in PowerDVD with those in my film.
In any case, should it be done be ear or by watching the amplitude of the signal?
thanks.

[Rick Reineke]

AFAIK, there are no real standard reference/peak/loudness levels for DVDs, as there are for broadcast submissions. Even major MP DVDs, the levels vary from one to the next. I normally set final levels for DVDs to peak around -5dBFS, unless the client requests otherwise. If a calibration tone is needed: -12dBFS.

(Usually for stereo audio broadcast submissions in the states -20dBFS ref. w/peaks not exceeding -10dB.. I think the ref. level in Canada is -18dBFS.)

[Stanley Szpala]

thanks. What work-flow would you recommend for the actual volume adjustment?

[Steve House]

The actual workflow is to use a calibrated monitoring environment. You need a couple of test files, a sinewave recorded at -20dBFS and a pink noise track also recorded with an average signal level of -20dBFS, and a sound pressure level meter, the $50 model from Radio Shack will do. You first load the sinewave into your editor, adjusting the track output level slider to unity gain, and adjust the master output volume slider on the editor's software mixer so the software output level meter reads -20dBFS. Then you switch to the pink noise file. Without touching the level sliders, the pink noise should playback hovering about -12 dBFS. Make a note where the sliders are so you can reset them to the same spot later. Now put the sound pressure meter where your head will be when you're mixing the tracks and working one channel at a time, play the pink noise file through your speakers. Adjust each speaker's volume control so the pink noise reads 82dBSPL on the sound pressure meter. (The target is slighlty different for TV broadcast, large auditorium theatre, or DVD home theatre but 82dBSPL is a good average.) Mark the volume control settings so you can always return them to the same spot. Now you have established a standard reference level so you can mix by ear to adjust volumes when the talent is whispering versus shouting or balancing music and effects with dialog, etc. When mixing you set the output sliders and the speaker volume controls to your pre-determined point and use the editor's mixer track level sliders to adjust the levels of each element.

Average dialog levels should fall somewhere around -22dBFS on an averaging meter in your editor, that would put it hovering around -14dBFS on the peak reading meters most NLEs use, with whispers and shouts correspondingly lower or higher respectively. Music and effects should be balanced in proportion to the dialog but the basic dialog level is what drives the level of the overall mix. Music only passages can go to higher average levels but music or effects peaks (gunshots, etc) should never exceed -3dBFS.

[Stanley Szpala]

this is super helpful! I've found a few www that sell the sin tone and pink noise.

A couple of question
1. For the sinwave, what frequence? 1kHz?
2. it's not clear to me why using the sound pressure meter instead of a software-type average sound level? I guess the volume should be adjusted "by ear" after establishing reference sound levels at the head location (using the sound pressure meter), instead of using the amplitude of the signal in the computer?

[Steve House]

Originally Posted by Stanley Szpala

this is super helpful! I've found a few www that sell the sin tone and pink noise.

You don't need to buy any test files. There are a number of places on the web where they're available for free download and you'' also find them on the CDs the accompany a lot of books on sound recording and editing. One free source is from the good folks that make Blue Sky monitor systems ... Blue Sky Blog Archive Blue Sky Calibration Test Files. I don't know anything about Vegas Movie Studio but Vegas Pro can generate them itself.

A couple of question
1. For the sinwave, what frequence? 1kHz?

Yes, 1kHz is the standard reference tone, same as you use for bars and tone in the program header.

2. it's not clear to me why using the sound pressure meter instead of a software-type average sound level? I guess the volume should be adjusted "by ear" after establishing reference sound levels at the head location (using the sound pressure meter), instead of using the amplitude of the signal in the computer?

Correct. You still need to use the meters in the editor to insure the baseline levels are correct and your peaks don't go too high but when you mix a program not everything is going to be at the same level. Two lovers whispering in bed are going to be at a much lower level than the machine gun rounds fired when their bank robbery goes wrong the next day. Music behind dialog will be at a low level so the dialog can be heard clearly but will swell in a dramatic crescendo when the hero walks off into the sunset. If you mix strictly by the meters you'll even out those dynamic differences in sound that should be a part of the storytelling. The overall level of dialog averaged out over the whole show should be about -22dBFS. That sets the baseline that everything else hinges on. But the meters don't know if the sound it's displaying at this instant is a whisper or a shout and obviously you want a whisper to be softer than the average speech and a shout to be louder. How MUCH softer or louder it should be is something you can only judge be ear. But to do THAT, you have to be listening in a calibrated environment. Otherwise when you hear something that is very soft, let's say, you have no way of knowing if it's soft because that's the way it is in mix or it's soft because someone turned down the speaker volume on your desk. The sound pressure meter is used for setting up the system's calibration, not for monitoring the mix.

[Stanley Szpala]

ok, I've got the dBC meter (Scosche SPL1000F), the sound files and I started the calibration.
I have a problem: I can't go to 80dBC without pain in my ears or at least discomfort. I think my hearing is hypersensitive, which could explain why I usually feel the volume in movie theaters is set too loud! Also I don't live in a desert, and I can't play anything that loud for an extended period of time (because people will kill me well before I finish this film)

I need to modify the protocol somehow.
70dBC is the most I can tolerate, but I probably will have to work at 60dBC of the pink noise (recorded at -20dBFS). So I'm not sure how to deal with this entire calibration protocol.

What do you think I should do?

I think I need to load to my Movie Studio some properly adjusted samples of dialog, music, sound effects, and look at the numbers. Is there a www with such?
thanks.

[Steve House]

The suggested levels for calibration are based on your hearing the soundtrack as you're mixing it at the same level the audience will hear it when it's played in their environment. Your average level of speech will not be blaring out at 80dB, the 80 only refers to the calibration pink noise. Large auditorium sound systems and motion picture studio dubbing stages are basically set up so that a -20dBFS pink noise will register 85dB at a position in the audience 40 feet back from screen centre or at the mix position on the studio stage. When you substitute normal speech at -22dBFS average it will not sound as loud. Since your mixing room is not a large theatre or studio dubbing stage, that 85 can be reduced to 80, 78, or 76 to give you the same subjective volume level. But if you calibrate to 60 or 70 dBSPL, you'll need to increase the level of the track in the mix to make it sound "normal" and the soundtrack will be way too loud when played for the theatre audience or when people play the DVD on their home theatre systems.

[Stanley Szpala]

I think I can "cheat" the protocol a bit without affecting the actual levels of sound in the final film:

Instead of playing the original pink noise file set at -20dBFS and adjust the volume until the sound meter reads 82dBC, I attenuate the pink noise by 10dB (by applying FX=volume, not the sliders in Vegas Movie Studio), and adjust the volume knob so that the sound pressure meter reads 72dBC instead of 82dBC.
I think mathematically it is correct, although slightly less accurate because signal to (ambient) noise ratio is slightly compromised.

I've tested it and it seems to be working, i.e. both procedures yield very similar position of the volume knob.

Anyway, I don't think I could safely calibrate my rear and center speakers at 82dBC, because they are low-power, and only my left and right speakers are of decent quality.

what do you think?

[Steve House]

I think those standards evolved for a reason and you deviate from them at your peril. You said you were going to submit this film for festivals, that means projection in a theatre if accepted. You need to have a setup where what YOU hear when you are mixing it is exactly what an audience member is going to hear when they are viewing it. You have no control over what their playback setting are going to be so you have to assume the theatre where it's going to be shown will be set up according to industry standards.

[Stanley Szpala]

it's hard to disagree with what you're saying. I'll make sure I comply with the standards in my final production. And Steve, I totally appreciate your advice!

[Stanley Szpala]

I went to a movie theater yesterday (RED), and I took the decibel meter with me to get a better feel for the sound levels. The dialogs were between 60dBC and 70dBC (actual readings, not max), with the lower end for "normal conversation". Music loudness strongly depended on the scene, but it was often around 80dBC and up. The loudest sounds (gun shots) were around 95dBC.

[Steve House]

Originally Posted by Stanley Szpala

I went to a movie theater yesterday (RED), and I took the decibel meter with me to get a better feel for the sound levels. The dialogs were between 60dBC and 70dBC (actual readings, not max), with the lower end for "normal conversation". Music loudness strongly depended on the scene, but it was often around 80dBC and up. The loudest sounds (gun shots) were around 95dBC.

Yep sounds about right... actually the very peakiest peaks might hit up to 105dBSPL (c-weighted, slow) on each channel. That's 85dB reference level +20dB headroom. Volume at the listener in a theatre when you combine the ouutputs of all surround channels can reach 120dBSPL momentarily. Remember the 85dB is a calibration level for -20dBFS pink noise ... it's NOT the level dialog will play at. Dialog is mixed at an overall average over the entire program of -22 to -24dBFS as read on an average reading meter, about 80dBSPL, and particular passages may be louder or softer than that depending on the mood of the scene. (You can find a lot of discussions on average program levels for broadcast if you google the the term "dialnorm.")