Best way to mix a soundtrack?

videoboy77 wrote on 8/11/2006, 8:55 AM

What is the best way to mix a basic video soundtrack with narration and background music? What db level should the whole soundtrack not exceed? What level should the bg music stay at while narration is playing? Should levels be set on the two tracks independently? I know this couyld get detailed, but im just interested in the basics. I'm trying to lock down a workflow for this and for years I've just been wingin it. Thanks!


flyerstl wrote on 8/11/2006, 9:07 AM
The only way to answer this is with non mathematics-specific words...


It's all about ears. If your narration is peaking (digitally, in Vegas) at aobut -12 to -6, then your music might end up being 15 db below that. But it's all relative. And there are many variable such as the apparent 'loudness' of the music track; whether your narration is a strong energetic read; whether you have no plugins or many dynamics plugins on the narration track; whether the music has heavy bass and percussion, or is light and airy.

Just use your ears. They are the best judge.
jmm in STL

jmm in stl

Windows10 with Vegas 11 Pro (most recent build). Intel Core i7-3770 @ 3.40GHz 3.90 GHz, 32GB ram, separate audio and video disks. Also Vegas 17 Pro on same system. GPU: NVDIA GeForce RTX 2060 SUPER. Dynamic RAM preview=OFF.

farss wrote on 8/11/2006, 4:25 PM
Beg, borrow or buy a copy of "Audio Postproduction for Digital Video" by Jay Rose and read it, several times.
A goldmine of information from a great guy.
As jmm says there's no magic solution but at least Jay's book will give you a skeleton to put some flesh on.

JeffreyPFisher wrote on 8/12/2006, 7:07 AM
To the level suggestions offered above, I would add EQ, too sometimes on the VO track to bring out speech with a boost 'round 2250 Hz 9again use your ears!) with the Track EQ. A subsequent cut in the same area on the music track with EQ can help things "sit" better.

Serena wrote on 9/6/2006, 4:45 PM
Amazon has only "Producing Great Sound for Digital Video" by Jay Rose. Is this the one you mean?
bStro wrote on 9/6/2006, 4:59 PM
Serena, isn't this it?

Serena wrote on 9/6/2006, 5:50 PM
Indeed -- thanks. Drew a blank when I searched -- must have mistyped.
farss wrote on 9/6/2006, 9:03 PM
Glad you found it, a goldmine of information.
And as I think I've said several time before Jay is a GREAT bloke. I don't remember now which forum he lurks in but last time I asked there I was kind of blown away by the amount of help he gave.

MH_Stevens wrote on 9/6/2006, 9:39 PM
A good book; I get it out at night when I can't sleep. But it's heavy and takes time. There are simpler overviews for a beginner out there. Try the J Rose onlin tutorials here:

Serena wrote on 9/6/2006, 10:10 PM
Thanks Mike for the pointer. Have to put myself out of the beginner class. Just haven't got a good text in the area. Have always made it it up as I go along, but sometimes (just sometimes) consulting an expert saves time and even gets a better result.
TorS wrote on 9/7/2006, 2:31 AM
Use your head, too.
By the time you are doing the mix, you will be quite familiar with what the talents are saying. You probably know every word by heart. But you will be mixing for someone who has never heard it before! Think about that, and lift the VO or dialogue just a little higher in the mix. Otherwise people may have a hard time making sense of it all on the first hearing.
Serena wrote on 9/7/2006, 5:35 AM
Tor, good points. Probably I should mention here that the original poster probably isn't reading this. I remembered that Bob had mentioned a text book in this thread and didn't find it in my search, so I bumped the thread back up to check the title. So if you're speaking for my benefit, then I know those issues rather well. Thanks anyway.
Bill Ravens wrote on 9/7/2006, 7:16 AM
There's a really great tool available for mastering a soundtrack, particularly if you include music in the soundtrack. It's here:
Jay Gladwell wrote on 9/7/2006, 10:48 AM

Hey, Bill, good to hear from you!

Excuse my ignornace (I don't consider myself an "audio person" per se), but how would Har-Bal work when mastering a sound track with more than music alone?

Bill Ravens wrote on 9/7/2006, 11:19 AM
Hi Jay..

Har-Bal is, fundamentally, an EQ shaper. If the vocal, including a v/o, was recorded in a room with poor acoustics, the recording will be pretty top heavy in harmonics that may make it difficult to hear, or even unpleasant to listen to. Some audio actually will tire a listener out. EQing audio will help this problem, har bal presents the frequency spectrum in a way the makes the job pretty simple. If you have several different vocals to bring together on a single audio track, harbal will help you match the volume level between each recording in a much better fashion than the native vegas audio editor. Harbal also simplifies the process of compression if the audio track is too dynamic. Unfortunately, HarBal won't work as a plugin to Vegas, so, you have to process your audio track seperately. This is a great tool for mastering at a fraction of the cost of something like Ozone, altho', admittedly, it won't do everything Ozone will.
technobaba wrote on 9/7/2006, 2:27 PM
Use the audio compressor just for fun. Leave "auto compensation" on and set the compression to 4x. Fool with the threshold.
Auto level compensation is horrible in general but a start so you see and play.

Rules of Thumb:
In my documentary style videos I adjust each track separately using the volume envelope.

Make sure multiple tracks do not add up to overflow, i.e. two channels added can peak out. Render the track (cntrl-M) and look at the waveform. You will see this visually.

Adjust so the final waveform is about half height on average. It should never peak past 100% (red on the level meter). If so adjust or use the compressor.

Fade the music down (volume envelope) when you have the narration.

Audio is an art but the above gets you started.

farss wrote on 9/7/2006, 2:50 PM
Ducking music under speech, hm, not the best idea.

A better idea is to change the Eq on the music track to leave room for the speech. If you've got music with a goodly amount of energy in the same spectrum as the speech you're going to have to push the music down a long way. At that point it may well cease to sound like music and start to sound more like a distraction to the speech.

The above depends on your choice of music, great sounding music is not necessarily the best for this use.

Or you can use SFPro's mood mapping to change the mix of the music to accomodate the narration, works quite nicely. I assume Cinescore can do the same thing.