Soundcheck went fine.
The singer ran the verse twice,
easy and relaxed,
and the meter sat
comfortably below the top.
You left the gain there.
Three takes later
she found it.
The real voice.
The one she only reaches
when she stops thinking,
on the last chorus,
on the note the whole song leans on.
Everyone in the control room
went quiet.
Then you played it back.
On the big note
there was a rasp,
a hard edge that was not in the room,
like paper tearing behind her.
The little light on the interface
had been red.
Nobody was looking at it.
Here is the mechanism.
A digital recording has a ceiling.
Zero on the meter
is the largest number
the converter can write.
Anything louder
is not written
as a bigger number.
It is written as the ceiling,
sample after sample,
until the wave drops back under it.
The round tops of the wave
are sliced flat.
A flat top
brings harsh new frequencies
that your ear hears as distortion.
And the part that was sliced off
is not hiding somewhere.
It was never recorded.
Turning the clip down
gives you a quieter copy
of the same flat top.
Repair tools guess at the missing curve,
and on a voice
you can usually hear them guessing.
Pulling the fader down
in the session
does nothing either.
The fader comes after the converter.
The damage happened before it.
So set the level
for the loudest moment,
not for the soundcheck.
Ask for the last chorus first,
at full commitment,
and set the gain on that.
With twenty four bit recording
you can afford to aim low.
Peaks around minus twelve
leave room for the surprise,
and the noise floor
sits far below anything
the room itself will add.
Turn a quiet take up later
and you lose almost nothing.
Turn a clipped take down
and you keep the damage.
Watch the light during the take,
not only before it.
The performance you are hoping for
is louder than the one you tested with.
– Serguey Asael Shinder
Top comments (0)