compressing dialogue?

[shilshoolon]

OK, I'm a real audio noob so here are some basic questions. I have some musical background(not much) so I can sort of use a compressor.

Here are some questions I was hoping you could help me with:

1)I read that level riding sounds better then compressing, but after riding the signal would you recommend some sort of light compression to flatten the "micro-dynamics" even more?

2)when compressing typical film dialogue, what sort of settings should I start from and why? I am not after numbers and ratio settings just the simple principle e.g. should I use the shortest attack time on the compressor to flatten the voice from peaks? or should I compress according to the voice like vocal compression?

3)Do you normally limit peaks with a compressor for soft limiting or a brick wall limiter?

I'll stop with the questions for now

Thanks,

Shil

[BrettMorello]

1) I always ride levels, and those dialogue tracks are being sent to a bus with some light compression. Always depends from what the sound is like (bad recorded reality or great recorded fiction or ADR, is it being mixed for TV/cinema/DVD).
2) see 1, depends on what is mixed for and what the recorded sound is like. Mixing a lot for TV, I find myself using very fast attack and release time, to be able to control my gain riding better and for a more snappy sound.
3)The brick wall limiter only comes at the very final stage, being applied to the final mix. Most of the time, with exception of commercials, the limiter only does that with my mixes: limiting the peaks.

[Geert van den Berg]

+1 on what Brett said!

[shilshoolon]

Great!! Thanks, that really clears things up now

Just wondering about the method of work you're mentioning in the first point. When working in film and using this method, do you level ride while the compressor is already on the bus or do you add it after the level riding?

Thanks for your help,

Shil

[jahtao]

For you, as you're not sure, it might be safer to ride faders with no compression on the bus at first, and then once your satisfied its all leveled out nicely, slap the compressor on and have another pass. Knock a couple/few dB's off with some soft knee 2:1.

[BrettMorello]

Originally Posted by jahtao

For you, as you're not sure, it might be safer to ride faders with no compression on the bus at first, and then once your satisfied its all leveled out nicely, slap the compressor on and have another pass. Knock a couple/few dB's off with some soft knee 2:1.

I always do the opposite. You can never 100% predict how the compressor is going to react on your gain riding, and how the balance with the rest of your mix will be! So with me the compressor is allready on, in an open plugin window to keep eyes and ears open that is used correctely. If needed, some peices are automated.

[shilshoolon]

Thanks thumbsup What would you recommend for this type of gain riding (lets say badly recorded dialogue) a single band or multi band compressor?

[jahtao]

Yeah Brett, I'm with you on this one BUT for Shil I thought it might be better to keep things simple as he's quite inexperienced. Don't want him overcompressing everything.

Shil, single band! Use automated EQ, noise reduction and volume rides to make it sound as good as you can. Any thing that sounds unacceptably bad should be replaced by re-recorded dialog (ADR etc.) ideally.

Does anyone here frequently use multiband for dialog as standard?

matt

[Dunebuggy]

I use 5bands DB max compressor for LtRt and very lite on the bands. For dial. tracks i use C1, a straight compressor with an EQ before.

[shilshoolon]

Thanks for the replies everyone

Originally Posted by Dunebuggy

I use 5bands DB max compressor for LtRt and very lite on the bands. For dial. tracks i use C1, a straight compressor with an EQ before.

Do you use the EQ before the compressor to cut out the low frequencies and some high or just lows? If so, what do you recommend to cut or not to cut in dialogue for film?

[Dunebuggy]

roll off both ends :120 hz soft slope and 10 000 hz soft slope , of course those parameters are automated and they move , sometimes a lot , plus the 2 middle ones are fully parametric and they move a lot too during a mix.I way more often cut than boost

[yazoo]

Originally Posted by BrettMorello

1)
2) see 1, depends on what is mixed for and what the recorded sound is like. Mixing a lot for TV, I find myself using very fast attack and release time, to be able to control my gain riding better and for a more snappy sound.

do you get more "snappy sound" by having a fast attack and fast release for me it sounds like you take away the fast peaks like a limiter approach?


For me snappy sound is when you have long attack and short release.

help me here!

regards benny

[shilshoolon]

Originally Posted by Dunebuggy

roll off both ends :120 hz soft slope and 10 000 hz soft slope , of course those parameters are automated and they move , sometimes a lot , plus the 2 middle ones are fully parametric and they move a lot too during a mix.I way more often cut than boost

Thanksthumbsup I did what you said and it sounds great.

I'm not sure if I'm cutting too much. I set the Logic eq to 120hz with a 6db slope on the low cut filter(the softest slope on the eq) but it seems that the slope actually starts to drop gradually from about 400hz . Is that how it's done or should I set it lower so it visually drops from 120hz? The same thing happens with the high frequencies cut.

[Dunebuggy]

To see if you re cutting too much for roll offs do a A/B test:

Eq in / Eq bypass

And see how bad you re affecting the quality of the voice ,you ll see that the voice range spectrum on production sound (boom , wireless) is sometimes pretty narrow.

It s always a case by case scenario , you might have too steep up the slope on youre HP LP roll offs.

The EQsettings that i mentionned are starters

[BrettMorello]

Originally Posted by yazoo

do you get more "snappy sound" by having a fast attack and fast release for me it sounds like you take away the fast peaks like a limiter approach?


For me snappy sound is when you have long attack and short release.

help me here!

regards benny

mm, maybe you have a point that snappy isn't the right word.

Mixing mainly for TV, the small dynamic range gives you limited option. With the fast attack, I try, indeed a bit limiter wise, to control very unhandy peaks like small handclaps and background noise etc. Especially in reality shows, but also often in fiction, there are several of those peaks which are a pain in the ass when trying to control dynamics of dialogue. I know with the recent trim function in PT you can pretty handy draw out peaks manually and then ride the gain afterwards, but you can lose much time handling such peaks. A short attack gives me the impression to have more control over overall dynamics with difficult sound.

The short release is the most important idd soundwise imo on many dialogues, although it depends on project.

[shilshoolon]

Thanks for helping. things are starting to sound much better

[bigshakes]

Hi Guys,

I mainly gained my 'Sound Tech' education through music but at this moment in time I am working in Audio Post for TV.

The problem I have is the way my colleagues use compression on a Fictional Soap Drama. Basically, I just don't see how it can work effectively.

They have a compressor set up at the end of the bus with set settings (no automation) that acts against all the dialogue tracks.

Now, if you are lucky this acts fine but thats more luck than judgement.

I would think that the way to approach it would be to take into account the following:

1) The Actor, 2) The Shot, 3) The Intended Performance, 4) Quality of Production Sound.

Surely it's not possible to compress all these varying signals, that change from shot-to-shot, scene-to-scene & take-to-take, with just a type of 'catch all compressor' because the signal level/sound is so dynamically different.

Now putting aside ADR, Actors Repeat Performance & awfully recorded material how would you go about compressing 3 actors in a scene that is made up of various shots, takes and intended level dynamics.

To me, it is easy to compress one shot of 'X' amount of people but to me that often changes once the scene moves on.

Apart from having individual cues spotted on different tracks for different treatment what is a more practical way of balancing and compressing out the dialogue to achieve a better dynamic range and gain structure overall for my mix.

Soz it's a lot
Bigshakes

[jahtao]

Bigshakes,

Have you tried their system for your self?

You don't say it sounds bad you just say you assume they must be wrong. I think its your preconceptions that are wrong here.

Matt

[Henchman]

Originally Posted by bigshakes

Hi Guys,

I mainly gained my 'Sound Tech' education through music but at this moment in time I am working in Audio Post for TV.

The problem I have is the way my colleagues use compression on a Fictional Soap Drama. Basically, I just don't see how it can work effectively.

They have a compressor set up at the end of the bus with set settings (no automation) that acts against all the dialogue tracks.

Now, if you are lucky this acts fine but thats more luck than judgement.

I would think that the way to approach it would be to take into account the following:

1) The Actor, 2) The Shot, 3) The Intended Performance, 4) Quality of Production Sound.

Surely it's not possible to compress all these varying signals, that change from shot-to-shot, scene-to-scene & take-to-take, with just a type of 'catch all compressor' because the signal level/sound is so dynamically different.

Now putting aside ADR, Actors Repeat Performance & awfully recorded material how would you go about compressing 3 actors in a scene that is made up of various shots, takes and intended level dynamics.

To me, it is easy to compress one shot of 'X' amount of people but to me that often changes once the scene moves on.

Apart from having individual cues spotted on different tracks for different treatment what is a more practical way of balancing and compressing out the dialogue to achieve a better dynamic range and gain structure overall for my mix.

Soz it's a lot
Bigshakes

Actually, this is exactly how I do it as well.
I don't use a compressor on each track, as one does in kusic, but ride the dialog into the compreossor.
It works much better. The job is to match the various actors eq, noise and level wise before the compressor.
And I use the compressor to smooth it all out a bit, and to catch peaks.

So, what the other guys are doing is in fact correct, IMO. And luck has nothing to do with it.

Mixing for film and TV is nothing like mixing music.

[Brent Hahn]

I don't use a compressor on each track, as one does in kusic, but ride the dialog into the compreossor. It works much better.

Yup. If you compress individual tracks, when a compressor on one track reduces the level on that track, the perceived leakage from the other tracks increases. Compress the buss, and the opposite occurs.

[JoeMilner]

OK, apparently I'm a contrarian, because I do the opposite: Filter, Compressor (light) and EQ on each dialog channel (in that order), and I ride the hell out of the faders.

And also, as my theatrical dialog stems have multiple centers (usually C-C-C-C-LCR), it's difficult to put anything across a "master dialog buss" as opposed to doing it in stereo, or LCR with a single center.

Just my 2¢

[danijel]

Originally Posted by Brent Hahn

Yup. If you compress individual tracks, when a compressor on one track reduces the level on that track, the perceived leakage from the other tracks increases. Compress the buss, and the opposite occurs.

I never noticed increased leakage as a problem (will try to listen for it), but I do find it a problem when one compressor does two people of which one is louder than the other. The other one actually gets sidechain-compressed so to say.
The other problem is volume buildup: if three people are shouting in one compressor, they will not get enough gain reduction, as opposed to when you have a comp for each of them.

Originally Posted by JoeMilner

Filter, Compressor (light) and EQ on each dialog channel (in that order), and I ride the hell out of the faders.

+1 on this - I like it better to do final EQ after compression for two reasons: 1. compression changes our sound color perception (by changing level) and 2. I somehow have the feeling of EQ-ing less when setup in this order.

[bigshakes]

Originally Posted by Henchman

Actually, this is exactly how I do it as well.
I don't use a compressor on each track, as one does in kusic, but ride the dialog into the compreossor.
It works much better. The job is to match the various actors eq, noise and level wise before the compressor.
And I use the compressor to smooth it all out a bit, and to catch peaks.

So, what the other guys are doing is in fact correct, IMO. And luck has nothing to do with it.

Mixing for film and TV is nothing like mixing music.

I'm fully aware that Mixing for TV isn't the same as mixing for Music. Give me some credit!

As people have stated out here, there tends to be a lot more going on in their dialogue than simply sticking it through a compressor regardless. Everyone seems to do a few stages previous to compressing. Whereas here all dialogue, no matter what, is on one track bussed to a compressor.

Just wanted to know about other peoples processes.
Cheers
Bigshakes