newbie question: Consistent sound with mutliple interviewees.

jrm27

Active member
Hi all,

I'm new to video creation, but I'm loving it and learning a ton as I go. I shoot everything with my t2i and an assortment of lenses. When doing sound, I've tried very hard to do as much second system audio as possible. I have a simple setup, but it seems to be working for me for now. I use a cheap little lav into a zoom h1 and then sync up in post with Plural Eyes. usually, if I'm doing an interview video, I only have one person to mic up so I can keep my audio pretty consistent through the process. However, this last video I did had 5 different interviewees. Some talk louder than others, some quieter, etc.. I tried to go through and manually set the volumes to be similar before I started recording each person, but it apparently didn't work to well.

After importing all my footage (I edited in Premiere), I tried to match audio levels from person to person as best as possible, but I'm still not satisfied with the result. I tried some compression, but using it in this application is still a mystery to me (hand me a compression pedal for an electric guitar and I'm more knowledgeable).

Anyways, any tips? Anywhere you would suggest I learn from? Thanks!

-jon
 
I would start with a more capable recorder, or a mixer in front of your current recorder, with proper level control. Any sound editing and clean up that needs to be done should be done in a DAW. I've tried to do basic audio work in Premier in the past, and found it extremely frustrating.
 
Compression isn't really what you're looking for. What compression does is squash the dynamic range of a segment, raising the level of the lower portions to closer to the higher. While you might use compression at some later stage, the first step would be to normalize your audio clips.
 
Thanks davefisk and Steve House, I really never thought of running them through my DAW first. Still learning my workflow!

Might either of you mind sharing your workflow in normalizing clips in a DAW? I've got a few different options between Pro Tools and Soundbooth.
 

That is a tutorial on calibrating your speakers/ system.

No what he was talking about is a plugin. In Protools it's called NORMALIZE and it's probably called that in everything else also.

It's kind of controversial in film, as in you won't find a lot of folks using it on narrative films. Gets used a lot on video, ENG and folks on a tight time line where acceptable and fast trump everything else.

What they do is take your sound clip and raise the peak to a given level.

The negatives are that the older ones always brought your peak to zero, which you don't want. Most now let you pick a level.

Because of this if you go clip by clip through your tracks and normalize them the peaks will all be at the same level. The downside of that is your noise floor is going to jump up and down because you have gained each clip differently.

You can do this to the raw files and you won't get the jumps but one loud smack, click or ??? will throw off the function because it's looking at the highest peak to use as a reference.

What I tend to use instead when I'm going to "normalize" something is use the GAIN plugin in Protools. With that you can examine the clip and see what the peak and RMS levels are first. Then you can decide if it's going to take too much gain to get up to the clip you want to match and you can compare the RMS value which is a better "loudness" indicator than the peak value is. It's slower than "normalizing" but you have more control and it's still faster than leveling by ear when speed is at a premium.

There are also a couple of "plugins", I'm putting that in quotes because I think at least one is an app/plugin combo, that will "auto mix" a track. You give it some parameters and it will go through and write the automation to even out the track. One is called Wave Rider.
Reports are mixed ;~) Some find it not worth it and other think it's the best thing since sliced bread.

I tried an early version and it was OK but kind of a pain to set up. Now it's at version 2 so maybe the kinks are worked out. I think WAVES has something like this also. The upside is it should be much smoother in the level changes. Much more "mixed" than processed, and still very fast.
 
Back
Top