EQ'ing for video/film

[Jon Fairhurst]

Here's a set of tips that I would like to share. The EQ part is based on infromation from a Jay Rose article, so I can't take credit for it. But I've used the technique, and have found it to be successful.

We face the problem that we want our films to have music, foley, effects and dialog, and we need the dialog to be clear, natural and understandable. We also want the music and effects to sound powerful, but not bury the on-screen voices. What to do?

Here's the deal:

Under 150 Hz
-------------
In this range you can pull down the dialog. It will help reduce plosives, handling noise and echo in large rooms. If the voices get too thin, lower the cutoff frequency.

For music, you can cut things like bass drums a bit at the lowest frequencies (say under 60Hz) to ensure that you have no sub-sonics, and to give you more headroom.

Low cuts are nice for safety, but if you're too aggressive, things get thin. Find the right balance.

150Hz - 300Hz
--------------
This is where the fundamentals of voice and many important instruments exist. I like to give the voices a slight boost and the music/effects/foley a slight cut here. If your voices are boomy, back it off. If thin, boost away.

One trick is to send your music to two sub-busses. When you're underscoring dialog, use the cut version. When standing alone, use the flat version.

300Hz - 600Hz
--------------
This region is less critical for voice. You can boost your music/etc here and get away with it. Cut the voices here to make room.

600Hz - 1200Hz
---------------
This region is critical for consonants. Boost the voices and cut the music. You lose some fast attacks on your music, but it's more important to understand the talent.

1200Hz - 2400Hz
----------------
This area isn't critical for voice. Cut the dialog, boost the music a bit.

2400Hz - 4800Hz
----------------
This region is important for distiguishing voices and instruments. Boost the voices and cut the music - especially if there are multiple voices. The downside is that your oboe will start to sound like a clarinet. You can push this range back up for the music when it stands alone.

4800Hz - 9600Hz
----------------
This is the *sizzle* region. I boost both the music and voice heavily here.

9600Hz and up
--------------
Don't worry about higher frequencies too much. It's mostly noise. If your tracks have HF noise, feel free to cut with a heavy hand.

The overall sound
-----------------
I mixed with this technique last weekend for the 48-hour film project. I didn't have time to think about it, so I just threw the above EQ rules at everything, modifying each band by +/-2dB to +/-4dB, being less agressive on the lows and more aggressive on the highs. You can actually be much more aggressive with the cuts and get away with it.

The voices sounded *great*. We used a long shotgun (our best mic) indoors, and the ambience was very natural. It didn't sound boomy, in a cave or anything like that. In fact, it sounded very intimate. It gave the feeling of being very close to the actors in a quiet invironment. You could hear lips, tongue and breath. There was just enough room effect to sound in-place.

The music sounded a bit hollow. This was great for my forelorn score, but wouldn't work in all cases. If this happens, just back off the EQ adjustments to taste. During the swells when there was no dialog, I wish I had gone with a more neutral EQ to give more power, but it wasn't in the time budget.

The bottom line is that I could run music peaks simultaneous with dialog and never lose an ounce of intelligibility.

The other "rules" I used were:
1) Mix nominal dialog so it peaks around -12 dB. Find the loudest normal speech, and set your fader for the -12dB peak goal (ignoring plosives and the odd loud sound).

2) When the dialog is too loud, use an envelope to cut only the offending syllables. This keeps your noise floor consistent between sounds.

3) Copy the dialog track and apply heavy noise reduction. Only mix this in when the dialog is too faint. Mixing this in won't affect the noise floor, and you've still got the unprocessed stuff from the main track in there, so the NR artifacts will remain somewhat hidden. You can be sloppy with the envelope, as there should be little noise between words. For dead syllables, mix up both tracks on just that syllable. (Fortunately, the EQ tricks help avoid dead syllables.)

4) Mix the music as needed. Rather than use compression, mix down the offending instrument or envelope down just the loud hits.

5) When done, especially if you are the composer, mix the overall music down another 2 to 3 dB. No one loves your score as much as you do. ;)

6) Review the whole thing. Envelope the music back up in the dynamic swells, if needed. Step 5 gave us more headroom - let's use it.

7) Check your peaks. The maximum should be about 1.5dB below full scale, just to be safe. You can peak higher on an explosions or gun shots (which are just a loud noises), but not as high on a loud voice or music, which have understandable information.

There are countless other tricks as well - especially if you have a multiband compressor on hand. But we're not mixing the world's loudest CD here. We're mixing for a film. We've got dynamic range. Use it.

In any case, if you're not a professional audio mixer, just go with the above rules, and your DVD print will have understandable dialog, safe levels, room for dynamics and a consistent level of background noise. The result should sound professional.

Enjoy!

[Tim Gray]

Originally Posted by Jon Fairhurst

The music sounded a bit hollow. This was great for my forelorn score, but wouldn't work in all cases. If this happens, just back off the EQ adjustments to taste. During the swells when there was no dialog, I wish I had gone with a more neutral EQ to give more power, but it wasn't in the time budget.

Some good info here for those starting to play around with sound. Question though... it sounds like you put 1 set of EQ on ALL of the audio. Unless I misunderstood the above passage, I'd EQ the dialog separate from the soundtrack (and everything else for that matter).

You have a really good bit about mixing for normal dialog peaks at -12db. Whatever point you pick (for a really dynamic track try -20db) it's important to monitor at the same volume. If you turn your monitors down a lot, you will start setting your levels more aggressively; if you crank them, you will start mixing too quietly in compensation.

I think the reference sound level used for film mixing is 85db. Get a little radio shack sound level meter and set your monitors at that level and then don't change them. If you don't have access to a meter, find a comforatble level and stick with it. Here's a good article on it:

http://www.digido.com/portal/pmodule...der_page_id=59

[Jon Fairhurst]

Hi Tim,

Great point about setting a consistent reference for your monitors. Walter Murch has written the same. You're in good company. :)

Personally, I haven't done that. I go with the -12dB thing, and set my monitors to sound comfortable with that level. That's partly because I mix at home. The level mostly depends on how much background noise I'm dealing with. :)

Regarding the "one EQ for everything", I sent my dialog tracks to one sub group and my music to another. I set one EQ curve for all of the dialog and a different curve for everything else. It's very efficient. Rather than lots of EQ curves on lots of tracks, I just had two to deal with.

On a past project I found one problem: different actors sometimes need different EQ. You can then EQ just the odd actor with whatever corrections you need to make, and still send everything through your master curve. The key is that you still put the music/foley effects into primarily different regions than you do the voice.

What amazes me is how aggressive you can get with band EQ as described above without the results sounding wrong. In fact, one approach is to really slam the EQ, listen and then back it off until you don't hear the EQ as problematic. The bottom line: if it sounds good, it is good!

[Michael Brady]

Quote:In any case, if you're not a professional audio mixer, just go with the above rules, and your DVD print will have understandable dialog, safe levels, room for dynamics and a consistent level of background noise. The result should sound professional.
Enjoy!__________________
Jon Fairhurst
http://PoorlyProjectedPictures.com Quote

Thanks to both of you for such free and helpful advice.I will be starting my filming project sometime in the spring of 2007. I am even more convinced that i will either hire a really good sound engineer/mixer for a couple of days to show me how to do things or work with one for a week free just to labour, carry stuff or whatever in return for some good practical guidance. There are some things you can practice with and improve as you go along. With sound i really get the impression if you are new to all this it would be cheaper, smarter and will save a lot of wasted time to get an experienced pro sound man to set show you the ground rules and some tips before you start doing it yourself. I have read Jay Rose's book on sound and found it really helpful. The forums are great for equipment and further advice but can you really beat the good pro giving you hands on advice from the very start?

Michael

[Dennis Khaye]

Very nice. Thanks for all these tips and links. This is one area I really need help in. Really appreciate it guys.

[Jon Fairhurst]

Originally Posted by Michael Brady

...but can you really beat the good pro giving you hands on advice from the very start?

If you can afford to hire an experienced sound-person/editor/mixer and can watch and learn, go for it!

I think there are three aspects to creating good sound:

1) The ability to hear things like compression, EQ, reverb, etc. I'll freely admit that I didn't have it at first. Not in the least. It took a few years of tweaking knobs and listening for the difference before it became second nature. If your sound guy says, "check this out. Do you hear the difference?", don't be afraid to say "no". Write down what he's doing, and try it out later until you get it. He might hear a half dB of difference. You might need 6 dB to hear it. Only after you "lock in" on the sound can you develop more sensitivity.

2) Technical understanding. I'm an old EE who chose DSP as my primary area of study, so this stuff is pretty natural to me. But it doesn't take a degree. Just dive in! Being able to picture and grasp what is happening technically is critical. Link that to what you hear, and you're most of the way there.

3) Creative tricks. Learn all of the tricks that the pros can offer. Then try your own things. Even crazy things. Then write down what you like and what you don't. Once you have a good toolkit, you can get a good sound very quickly. The bag 'o tricks won't replace items 1 and 2, but the tricks can save you time and get you close to what you want. Then use your technical knowledge and your ear to dial it in.

But, hey, I'm still learning. If anybody had any more tricks for this old dog, I'm more than happy to learn them!

[Michael Brady]

This thread has the makings to be something very useful indeed.

That was a great post Jon..... Those comments on the ability to hear things is so true. The CD disc that comes with Jay Rose's book demonstrates how important it is to listen. The different bit rate sampling going from 16 bit down to 8 bit was useful. The different wave forms associated with different instruments the same. I have to admit though that three quarters through the book it became pretty hard going to understand and way too technical for me at my stage of learning.

I know there's lots of other topics on sound elsewhere but has anyone got something to add to this thread based on the things alreadys mentioned?

Michael

[K. Forman]

I just found a book yesterday, Sound for Digital Video by Tomlinson Holman. I'm trying to learn all aspects of video, and sometimes get lost. I picked up this book, flipped it open to a random page, and started reading. Now, I've heard people mention Pads on mics, but never understood what it was. He explained it by comparing it to an ND filter on the lens-and it made perfect sense! I bought the book. He also goes into many of the dynamics of audio, levels, equipment, the whole enchillada.

Pretty soon, I'll understand most of what you sound guys are talking about ;)

[Tim Gray]

Originally Posted by Jon Fairhurst

Hi Tim,
Personally, I haven't done that. I go with the -12dB thing, and set my monitors to sound comfortable with that level. That's partly because I mix at home. The level mostly depends on how much background noise I'm dealing with. :)

That's currently what I'm doing too :) Do what I say, not what I do. I did notice though that monitoring at one level (a comfortable one) did wonders for the consistency of my work. I bet it was around 85 db anyway.

Originally Posted by Jon Fairhurst

Regarding the "one EQ for everything", I sent my dialog tracks to one sub group and my music to another. I set one EQ curve for all of the dialog and a different curve for everything else. It's very efficient. Rather than lots of EQ curves on lots of tracks, I just had two to deal with.

I figured I had misunderstood you there. Just wanted to be sure.

Originally Posted by Jon Fairhurst

On a past project I found one problem: different actors sometimes need different EQ. You can then EQ just the odd actor with whatever corrections you need to make, and still send everything through your master curve. The key is that you still put the music/foley effects into primarily different regions than you do the voice.

Yes, I just finished going through an hour long documentary with footage from many locations. The footage filmed early in the project had pretty poor sound (improperly set up wireless lavs -- wireless for no reason!) while the footage later in the project had much better sound (after I came onboard). I found it was ultimately best to split each subject onto their own track. Each voice got it's own EQ, some got compression, and then a lot of manual riding of the gain.

For the record, I know I was curious to know how well one could go from FCP to Logic to do a final mix. It can be done without too much difficulty, but I ended up not using the XML export (which if it worked, would have been really nice). Plain old OMF worked great. I then exported the finished mix as separate stems: voiceover/narration, interview audio, music, and environmental audio. These went back into FCP and everything turned out pretty good.

---

Those books seem to be a good start. I got my start doing live sound. I have a technical background too, so tech stuff does come pretty easy to me. John has it right though - dive in. You will need to learn some technical things since knowledge of your toolset is important to effectively use your tools. On the other hand, you can and will learn that as you go. You will develop an ear for this stuff as you work with it. I have learned far more by doing and learning than by reading books. The books are a good roadmap, but getting hands-on experience is what solidified a lot of this stuff for me.

One of the nice things about audio is you can "train" yourself all the time. Every time you listen to music on the stereo, dissect it a little bit. Try and figure out what you do or don't like about the sound. You can do the same watching movies; split your time analyzing the visuals with the audio.

---

John said this already, but this is a tip worth repeating:

You can be much more aggressive with cutting a frequency as opposed boosting it. The corollary to this is you should always try to cut instead of boosting.

Say for example you have a voice that is a little thin in the lowend. You might be better off cutting the midrange a bit and bring up the overall gain instead of adding a bunch of lows. Think subtractive instead of additive.

---

If some thing sounds 'wrong' with your audio, but you're not quite sure which frequency you need to adjust, add a relatively narrow boost with a ton of gain. Move the frequency around until you find the most offending frequency, and then change the boost to a modest cut. It's a good way to find problem frequencies that need addressing.

[John M. McCloskey]

Great info here, I wish we had a thread on DVinfo which pertained to using Music with editing and mainly editing styles using Foley, background music, scoring, Music Libraries, ETC. Really enjoyed the info from you guys.

[Jon Fairhurst]

I'm a regular at Northern Sounds. That forum targets sample libraries, samplers and synths from the composer POV. There are often threads there about mixing, mastering, monitors, levels and all that. Many of the composers there are writing for pictures and games.

I'm really amazed that there are so few people from the video-side checking out the composer's world and so few composers checking out the visual production and editing world.

Maybe that helps explain why the film guys often value music so little for their productions. ("I have a $100K budget, but no money for music.") The composers haven't been here infusing the video world with their values.

[John M. McCloskey]

Jon thanks for the info, I agree with you about the lack of info in theory of video editing and composing. It seems like most of the video related info we here of is the tangable, technical aspects of what buttons do what and workflow and also hardware interests with the entire production process. It would be very interesting to here the psychological tips and tricks you can use in editing and composing which makes this art we all work with more dramatic and meaningful. With theory nobodies right and nobody is wrong but putting individuals ideas of how and why they score a show the way they do opens up people to try something new. New ideas lead to new ideas.. Thanks

[Jon Fairhurst]

Originally Posted by John M. McCloskey

...It would be very interesting to here the psychological tips and tricks you can use in editing and composing which makes this art we all work with more dramatic and meaningful...

For the 48-hour film project we were assigned historical fiction. We pointed the lens at a mythical interrogator from the red scare of the 1950s.

I opened with a low bass drum and C2 piano hit on each title transition. (There were quiet notes in between the hits.) I repeated that theme at every stress point during the interrogation with varying intensity, until it peaks.

The final scene is a discussion with the main character about the morality of these interrogations, and the music is forelorn and melancholy. The discussion turns to argument and the music builds. When the interrogator effectively has his back against the wall, he snaps into his aggressive interrogation role. From silence the piano and bass drum give one final, dramatic hit, amplifying the abrupt change from discussion to inquisition. It's like a gut punch, cementing the emotional link to the earlier scene.

Neither picture nor music alone are all that powerful. Put them together and it can be transcendent.

[Jon Fairhurst]

We just got the news: our 48-hour film project made it into the second round.

Initially there were three groups. Ours (Group A) had 12 entries - all of which were turned in on time. Seven from Group A made it to the next level, along with five others.

The "Best of Portland" will be screened at 7pm at the Hollywood Theater...

[Patomakarn Nitanontawat]

Another tip I would like to chime in, a good investment would be to purchase some denoising plugins, either soundsoap, waves, or if you've got dough to spend, the Cedar lines. 60 percent of the dialogue captured on set has either a generator humming in the background, or the producer sipping his coffee. You might not hear it on the headphones, but played through the biggy cinema systems you will definitely hear it.