I'm certainly talking in terms of EQ. Also, I find it important to specify that embellishing voice (dialogue, VO, whichever) is a premixing duty. I usually stick with Pro Tools' RTAS EQ3 7-band EQ which is the most advanced one I have. Here's how I go about it:

I usually look for roundness around 200/300Hz, I-can't-remember-the-name around 3/4kHz and presence around 6/10kHz. I grab a maximum Q/maximum gain point and scan the whole spectrum while previewing the output. On most of the spectrum's width it'll add an unpleasant touch, but there are three points where it sounds less worse and that's how I find my spots. I then lower the gain to +6dB, my Q to 1.5 or 2, and fine-tune the gain, usually around +2 or +3dB. I'll regularly bypass the effect to ensure I'm not making it sound worse.

What way do you proceed? Talk about anything, I read a lot about the aural exciter, you might want to add reverb, a tiny delay to fatten it up, do you ever need to roll out some freqs?... whatever floats your boat, please feel free to share :)

3 Answers 3


i do very little EQ. its the basic frequencies of a particular voice i treat. it varies from voice to voice. i do dialogue for films. so i need EQs but with extreme care. the mids and lows( to give the "body" to the voice) are the most important. presence will be given according to the perspective and the nature of the shot and the space. thanks.

  • Great answer. I wonder how you treat severely noisy or dull recordings? What reconstructive tools do you have and use?
    – Utopia
    Commented Aug 13, 2010 at 20:11

6dB!?! GOOD GOD MAN! lol

That is a lot of EQ to bring to bear on a voice track. I can think of occasions where that might be required, but that's usually on a music mix of some sort where you're dealing with a much denser soundfield.

I usually start out by getting rid of the real loud mouth noises, and then I'll go on to EQ. I have some nice speakers at work (Blue Sky), which have a nice and flat response. I just go through and try to even it out. People are very practiced at listening to dialogue, so I prefer to just keep things "natural sounding." Whatever is required to get it there.

After that, it's usually some very gentle compression to bring out the body of the voice, and I'll leave it alone after that until I get into the actual mix. If I find I'm having any conflicts, I'll do everything I can to leave the dialogue alone (unless of course the piece calls for it to be subjugated). So, most of my fixing at that point occurs in other tracks, to preserve the dialogue as best I can.

  • What do you use for de-essing? Do you do any de-essing?
    – Utopia
    Commented Aug 10, 2010 at 0:51
  • And, how would you go about handling a peaky, tiny, harsh and no-bass voice like that of Spongebob Squarepants?
    – Utopia
    Commented Aug 10, 2010 at 0:53
  • @Ryan - hmm forgot about de-essing. lol i do that at the mix stage, because i prefer to see if it's going to be masked at all by anything else in the mix. if there's reasonable masking going on, i'll leave it, otherwise i'll slap a plugin on the problem dialouge track and automate that sucker. i'd handle spongebob the same way. i'm not trying to make the dialogue tracks "flat" just "smooth and even." sorry, i don't really know how to describe it beyond that. i just trust my ears and their gut reaction. Commented Aug 10, 2010 at 1:01
  • @Shaun I do not EQ with +6 of gain, it's only an intermediary step. I usually end up gaining specific spots at +2 or 3dB. I edited my first post to specify this ;) Commented Aug 10, 2010 at 1:22
  • 1
    I believe in Justin.. 6db in some EQ's it's a lot.. but maybe not in digidesign ones.. They're really tough.. And don't ask me why.. but I rather cut than boost.. sound better to my ears.. :) And if you allow me to.. Massey de-esser it's great.. you have a full functional demo in his site! Try it.. the invert phase option works great in some specific situations. :) Commented Aug 10, 2010 at 21:49

Assuming the recording and levels are good in the first place. I will almost always roll off lows and shelf the highs, even if its just a little. Human voices don't really have much below 120hz or above 12khz so get rid of it. Next I work low to high on the frequency band.

It usually looks something like this:

  • A lot of voices often have a specific annoying tone in the 100 - 800hz range. If there is one, I'll do a narrow band cut there.
  • Next I do a wider band adjustment of the "boom" in the lows. A thinner voice I will boost, and a boomer voice I will cut.
  • Some voices, females in particular sound nice in the 600 - 1000hz range. If its there, I'll boost it a bit.
  • There is "presence" in the 1 - 3khz range. If the voice needs to be more dominant or sound more forward, you can boost. The opposite is true as well. Just be careful here, too much in this range gets annoying and can give people a headache.
  • 3 - 5khz has some interest effects on the voice. Its near the "presence" but not high enough to effect hard consonants. Sometimes I boost, sometimes I cut, usually I don't touch this.
  • 5 - 12khz and on is where consonants come in. If the person isn't enunciating, a boost here may help. Cut harsher voices. Its where the "f", "t", "k", "s" and "p" type sounds show up. Sometimes there are narrow bands that need to get cut here as well. It can also act as a psuedo-desser when you cut here.
  • Anything above 12khz isn't naturally a part of most voices. There are some over tones that affect the sound but only minutely. You can usually cut this out entirely to help fit it into a busy mix. Like I said, I'll usually do a shelf filter to -6db or so.

After that a compressor with 2:1 or 4:1 ratio will usually sound nice. Maybe reverb/delay, but that's usually during the mixing phase.

I only do all of this if I have time or it must be perfect. Realistically when I'm on a time crunch, do a high pass at 120hz, wide band adjustment in 100 - 500, presence adjustment in 3-5khz, compress and call it good.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.