Sound Design Stack Exchange is a question and answer site for sound engineers, producers, editors, and enthusiasts. It only takes a minute to sign up.
Sign up to join this community
Can this be a thread for tips on mixing dialogue?
Do you EQ the voice a certain way or ever use different sync for different characters? Reverb/delay?
Do you have a certain workflow for the dial mix?
Do you treat the mix differently for things like camera angles or whether the character's back is facing the camera?
Thanks!
Ok here are my 2 cents from a little bit of experience in dialogue editing/mixing:
1) Whistles must be edited out by the editor and fixed before it hits the mixing stage.
I have mixed narration and dialogue with many whistles, and when I was starting out I tried to find the exact frequency of the whistle (2.5 to 4K) to notch out or de-ess with a dynamic notch EQ or Oxford Suppressor, what have you, and the amount of time I spent trying to fiddle with de-essers and parameters to make it breathe well with the rest of the voice, it would have saved me 80% of the time if I just snipped out the whistle. I can see them now in waveforms and I have a CH and S sound copied at the end or beginning of the session to fly in if I come across a bad whistle. It sounds so much better and doesn't give you unnecessary processing on the rest of the voice if all you're trying to handle is 2 or 3 whistles in one minute of dialogue or narration.
2) EQing a voice is different for every voice. I have heard mixers tell me "I ALWAYS add 'air' to the voice around 16K and notch out the 2.5K automatically" etc., and this is utter rubbish. Every recording is different, different acoustics, different mics, different recording equipment being used. Each voice you mix will require you to first hear it newly and EQ it specifically for that voice. I have never come across 2 different recordings having the exact same EQ. You have to listen.
3) Absolutely run low-pass filters on a voice if his back is to the camera. Your reference is any Pixar film (specifically The Incredibles) when Samuel L. J's character is searching his house for his super-suit, and his back is to the camera, and his voice is muffled, and he turns to the camera to say "Woman! Where is mah supah suit!!!" his voice is all of a sudden more present and brighter - it adds so much more dimension to the mix when the voice is mixed like that. It's a volume-ride and low-pass filter + reverb ride type of thing.
4) Something I learned from music mixing but also applies to final mixing, which is you EQ the voice with everything in the soundtrack playing - don't EQ it solo. You can set a temp EQ with it in solo but always check how it sits in the mix with all the other elements.
5) Adding early reflections to a voice can add a lot of realism and presence to a voice that's not possible with EQ.
Depending on the situation handed to me, my role and the quality of the recordings, I take different approaches.
In an idea situation: The sound recordist may have recorded onto multiple mics, lapel + boom, so I check which is sounding best first.
If the sound recordist is good, thye will have taken roomtone/presence/atmos, so I'll just cut the dialogue as tightly as possible with in/out fades, use noise reduction/deesser/declick if needed and set levels nice and evenly so they hit where I want them to, Lapel mics will need some EQing to match the boom. I then add the room tone and edit the levels so the edits/cuts are not noticeable.
If the sound recordist is very good they will have also taken wild tracks of dialogue and then I can replace any lines which I may have discrepancies with. If the sound recordist is further still on the ball they will have also recorded some foley in the environment which I can then continue to add in so I can make sure I don't need to use any production track other than for dialogue.
Then I will consider recommending doing extra foley or ADR, at which point I hope the Sound recordist is a god and has taken some convolution recordings so we can match any ADR to the scenes.
In a non idea situation: it's a case of a hell of a lot more work than should be needed to find a few seconds of atmos in takes to loop, noise reduction, filling gaps with what you can, spending ages with EQ and reverb to match ADR to production. So not needed for an extra few minutes on set to record.
I am believing more and more that mixing dialog for film/TV may be the most difficult and under appreciated of all the crafts in the post-production process. There are no rules, no silver bullets or do-all plugins that can do the job for you; it all boils down to having well trained ears, lots of patience and tons of experience.
That said, there are a few procedures that most dialog mixers I know will run through:
Hope that begins to address your question!
Just like everyone else I first start by going through and smoothing all the transitions out, picking mics, and inverting phase on problem mics. Sometimes while editing I'll even run some RX to see if I can get room tones to work. This all depends on what type of film it is. A horror film with lots of sound design allows you to get away with more than say a family drama with sparse music and sound effects. Once I've done my edit pass and finished up RX cleanup, I'll set up my mix pending on what type of project it is and what its' dialog deliverable levels are. Pretty much if your final product isn't mixed to the specs of the distributor your mix is going to get screwed up or bounced back for you to fix. Feature work allows for more dynamics than television work so I'll talk about that first. Pretty much you start by calibrating your room. Solo each speaker and play Dolby Pink noise to the volume of 79Db on your SPL meter with it sitting around the position where your head normally resides. The Spl meter needs to weighted to C level and set it to slow. With features I usually start with a gentle compression, somewhere around 2:1 to 3:1, on each dialog channel. After that I run it into an automatable 7 band eq. After that it goes into a Dialog bus to which I usually throw a limiter on. That limiter is set almost completely open, it's just to catch the occasional clip. The Dialog bus gets routed to the main bus to get routed to the final mix. For the verbs I usually run 2 separate ones, a close verb and a far verb. These are just Dx verbs that each channel gets routed to, then these verbs get routed into the dialog bus. Once that is set, I then go in and snap shot reverb presets pending on the locations in the picture. If I'm using something like TL space that can't change impulses, I'll usually use multiple ones for the different locations and set up the extra busses. After I'm ready to start my mix, I'll listen with my ears to get a good mix. If I'm ever sketchy on where I'm sitting at I'll nab out the SPL meter, c weighted to check that my dialog is somewhere between 70Db to 90Db. I don't really have it out that much though. The meter usually comes out a little at the beginning of the mix and at the end of long days. As far as Eq'ing. For features and shorts it's all about clarity of the dialog. I'm always using eq for matching microphones, notching out noise (sometimes after RX there's a little harmonic ringing in there you gotta notch), and for changing perspective. Around 1k is always a good spot for perspective trickery. Take some away to lose presence. The real tricky stuff is when you have noises that happen on set that change over time and you have to automate the notched frequency to match with the sound. It can get pretty complex with Eq'ing, but I do usually add a little top end. As far as the Verb, I get something to match the scene, which sometimes means using a predelay, and throw everything to it just a little bit to put everything in the same environment. After that you just mix and the rule is that Dialog is Key! Generally deliverables on films are -27 Db Rms for Dialog and -6db peak level.
For Television it's a whole other beast. Pretty much you're trying to compete with the washer and dryer going off while baby screams bloody murder because you forgot about your tea on the stove! The routing is all pretty much the same except now you have to get your dialog uber loud with a low peak volume. -25db Rms for Dialog and a -9 Peak level is a loud one. Sharp peaks, doors, cell phones in production audio all become your worst enemy! Thankfully we have Limiters. I'll go a little harder on limiting and compression all the way around and I'll add an extra limiter to the bottom of every dialog channel strip. Then I'll keep messing with the levels of all the limiters and compressors until the final output meets what I'm looking for. Once you have that it's smooth sailing. Just mix with your ears! One note on the limiters at this stage is that you really want to go a little heavier on the individual tracks rather than slamming everything into one limiter that has to handle the dynamics of all the mixed material. Limiters handle better with less information to deal with. Another thing about the EQ for the dialog is that you generally want to cut out a lot of the lows and add highs. Cut a lot of 200hz and add a shelf above 2kz.
I hope this sheds some light on it. I'm sure everyone has there own way, but so far this has worked for me.
Just found out about this forum board. This is really cool! Thanks to everyone for being here!
Utopia, nice answer and some really interesting tips. I too am always searching for new and improved ways of dealing with dialogue and ADR. I did post a question a while back on SSD and got soem terrific feedback. Here it is: Processing ADR to match interiors and exteriors
I feel that when dealing with dialogue in post, it is subject to a series of stages whereby you clean and sync it and then with the ambiance start to EQ it to match the space, perspective, camera angles etc. From there, there will be obvious tweaking and leveling to make sure that your dialogue doesn't get lost in the design.
Good luck.
