The story behind it:
Two days ago I did a “music production using free software” workshop at the Grazer Linuxtage (in Graz, Austria). My plan was to go through the process I usually do for my songs and to do the vocal recording there in a kind of a “live” situation.
For time reasons, I already prepared the drums (drumgizmo / CrocellKit), bass, and guitars (a Les Paul recorded “direct in” + guitarix) upfront. The vocal recordings were only two full takes, where I used the first take “as is” for the main voice and the second take for doubling at some specific places. Of course, I ran into beginner’s mistake #3 twice - some clipping during the last chorus.
However: given the circumstances, I think she did well and the final result doesn’t sound that bad at all.
And here’s the final version, where I polished up mainly the vocals:
What I did compared to the previous version:
better cutting of the best snippets from both takes
manual snippet-by-snippet region gain adaptions to get a better overall balance already before hitting the compressor, while keeping the dynamics of the vocal performance within each snippet.
a quite sharp low-cut filter (LSP), because something in the room caused a permanent rumbling in the low end.
further EQ (LSP), compressor (LSP), and de-essing (Calf)
multi-band compression around the 1k-3k band to cut down the “harshness” of the sound when her vocals become “loud”.
a few resonance cuts (LSP) with automation to be only active when needed (2 places with 2 cuts each)
