I recently tracked a female vocalist with a large dynamic range and horrible mic technique. After trying lots of the usual compression and limiting schemes, I have finally resorted to going through each song and manually editing the volume profile, to the point of editing individual words. This is such a pain, and I am convinced that there has to be a better way to do this. This is an hourly-paying gig, but there are limits on my patience, and her pocketbook!
SOS Forum Post
Technical Editor Hugh Robjohns replies: With a combination of poor vocal and mic technique you are really in the mire. Levelling out the signal dynamics is only part of the problem. When a singer moves around in front of the mic, it not only varies the signal level, but also the tonal quality (due to varying amounts of bass-boosting proximity effect and the shape of the mic's pickup pattern), and possibly even the perspective of the signal (the ratio of direct sound to room sound). Reducing the excessive signal dynamics by whatever technique may well simply exacerbate these other problems.
However, getting back to the dynamics, a compressor can only react fractionally after the event, and the transition from the linear working area to the compressed area will tend to draw attention to the processsing itself, even with a soft-knee compressor. I'm assuming we are talking about rock vocals here, but you might find you get better results if, rather than using a 3:1 or 5:1 ratio with a fairly high threshold, you use a low ratio (1.5:1 or maybe even 2:1), with a very low threshold. The idea here is that the entire vocal dynamic is being compressed all the time, rather than just the peaks. Using automatic, or programme-related, attack and release times if your compressor has the option will also give better results than a fixed release-time system. However, it still won't be great.
Riding the gain manually can produce better results than a compressor, largely because you can predict the dynamics and react in a smoother and more aesthetic way. It's a skill that takes time and practice to master, and often a rehearsal or two to fine-tune. In your case, as the recording has already been made (and assuming the entire dynamic range is clean), you should be able to use your DAW automation very effectively. Play the vocal track and ride the fader to try to balance the level as best you can, recording the moves as automation changes. You can then go back and edit the automation data to tidy up your moves, thereby saving a lot of practice time, to produce the best practical result.