Interviewee starts loud but gets quiet…Posted by Frank Miller on February 14, 2023 at 11:05 pm
This is something I encounter a lot.
Someone who is not used to being on camera will begin a statement or an answer to a question very loudly, and then taper off as they get to the end of the sentence.
Apart from setting a keyframe at the beginning and end of each statement and ramping up the volume, is there another way to fix this?
February 15, 2023 at 3:08 pm
This is what audio “Compressors” are for. Simplified, an audio compressor takes audio above a certain level, the “threshold,” and compresses it down a certain amount, the “ratio,” decreasing the dynamic range of the part above the threshold. You are squishing the louder parts and the quieter parts closer together. You then add “gain” back to raise all levels up together without over modulating. This is all done within the compressor. I am not familiar with Resolve’s options for audio compression. Compressors can get complicated and generally include a limiter, but the idea is simple. Decrease the dynamic range of certain parts of the audio signal so everything is compressed together, allowing you to raise all levels together without overmodulating the loud parts.
February 17, 2023 at 12:49 am
Four things that help. I do the following. Compressor limiter on the channel. Automate fader moves to chase dialog levels. Use the new dialog lev TrackFX on the channel to do an automatic version of chasing manually using automation. Finally keyframe within the clip. Personally I do steps one and two and occasionally step three as well. I have used clip based keyframes but I prefer to ride a fader with automation as I have the Fairlight mix panel.
On the Fairlight page you might not see the Track FX so go to the three dots on the top right of the mixer and tick them so you can see them and activate them
March 6, 2023 at 7:57 pm
There is / was freeware called CN Levelator, which I don’t think has been updated in about 10 years, but I still have it on my PC (Windows 10). You simply right click, and send your source file to Levelator and it evens everything out as the name suggests…afterwards I’ll run my file through Izotope to refine it even further. Results have been quite good.
Sorry to say that I tried to install it on a new Mac and I got an error message that the developer needed to update the app, so if your on a Mac it may not be an option.
March 8, 2023 at 12:05 pm
I’ve rarely noticed this problem in many years of interviewing. Maybe you are more aware of it than I. Simply adjusting the levels during the edit works for me.
I’d be very curious, though, how other people have handled this during the actual interview. Modeling a quiet tone of voice might help. Or simply telling the interviewee that the microphone is very sensitive, and please, it would be nice to have this as conversational as possible.
March 8, 2023 at 12:58 pm
I would not attempt to change the interviewee’s delivery unless it is too quiet in an absolute sense, meaning it can’t properly be recorded. Any reduction in dynamic range should be done after the fact if possible.