[LAU] Looking for voice recording suggestions

[hollundertee at gmx.net]

On Sat, 15 Dec 2018 12:41:56 +0100
Fons Adriaensen <fons at linuxaudio.org> wrote:

> On Fri, Dec 14, 2018 at 11:54:14PM +0100, hollundertee at gmx.net wrote:
>  
>  [...]  
> 
> Recording voices well can be a lot more demanding than most
> inexperienced sound engineers expect.
> 
> Some hints (probably too late...)

Indeed to late for the recordings, but thank you anyway. I'll still
have to do all the editing tomorrow morning and there will always be a
next time. 

I made a whole bunch of mistakes along the way, the most stupid one
first: I had someone whistle a melody and gave him said melody on my
headphones (AKG702). The melody was much too loud and bled into the
recording. These phones are probably not idea for recording, but the
main issue was definitely the volume. Good thing the whistling wasn't
usable anyway as it was rather out of tune and inconsistent. I think
having three or so people whistle live would work better, but we
decided on a different music track as fallback.

> You probably want as little room sound as possible, unless this
> is 'radio drama' and the room matches the scene to be recorded.
> So go for a dry room, and use whatever damping material you can
> find. Put it where the mic 'sees' it, e.g. behind the actors.

I put a blanket in a wrong spot. Well, there was pretty much only one
wall where I could fixate a duvet. The room is roughly square, I had
the mic rouhly center of one wall in ~2 m distance, facing the opposing
corner to the left. There's a large box containing some network
hardware or somesuch in that corner. The duvet was on the left side
wall. On the right side wall is a front of windows. Turns out some
people turned the mic to face the opposing wall rather than the corner
without me noticing. I don't know why, I guess they like symmetry or
right angles. I think the reverb wasn't too bad though.

> Background noises can be a real pain in case you need any editing,
> so find a quiet place.

That worked out well, no issues with outside noise.

> Be conservative with levels while recording, keep your RMS level at
> around -20 dB FS, in particular with non-professional speakers.
> Higher recording levels will NOT give you a better S/N ratio, and
> you can always adjust levels and reduce peaks later. Few things
> will upset you actors more than having to repeat a recording just
> because your signal clipped.

I probably recorded some people with a too high level, but the speakers
were vastly different. From large men with powerful low voices to tiny
women with whimsical voices. I didn't adjust my levels a whole lot
between those and used whatever metering Ardour has set as default.
I did have the limiter of the UA-25 enabled and there were a handful of
cases where it was active.

> If actors are reading from paper, have them hold it at the side of
> the mic, not below (that would make them look down and change their
> voice). That way they not speak directly at the mic and that will
> help avoid pop noises.

That sounds like a very useful tip. A lo of them had their lines on the
side but down as well. Funny enough most also read from the paper when
it was literally a single word.

> As already suggested, don't let the actors fiddle with the mic.
> Be strict about that.

I thought they didn't touch it but it seems some of them did without m
noticing.

> If you have highpass filters, use them while recording if possible.
> Cut out everything below 100 Hz.

I didn't apply any effects during recording. Does it matter whether it
is done during the recording or in post?

> For post-production you'll want at least a peak limiter in your
> master strip, and probably some EQ on each track.
> 
> Good luck !

Thank you very much. I don't know whether I'm any good with a EQ, but I
will try.

We cut down on our voicelines but a number of other teams asked whether
I could record them, so I ended up doing hardly anything else today.
I didn't find a good workflow in ardour, but it somewhat worked.
For our own voices recordings I used the playlist feature. Not sure
that was a smart thing. Exporting certainly seems easier with separate
tracks.

So that's what I did for the other teams, record on a track per line.
Sometimes a track per take, sometimes just keep recording those
multiple takes in a single clip. I then stem-exportet them, because
that seemed to be the simplest way to get a wav file per track. The
padding at the end is rather unnecessary though.

Because I needed to add new tracks I tried to add multiple at once and
got silence rather a recording on some of them. That seems like a bug
in Ardour. When you add tracks during a session every odd one is
non-functional. There may be more conditions to that, but it certainly
was annoying having to add twice as many tracks and then remove half of
them to get usable ones.

So the lack of knowledge on how to properly track and organise using
Ardour slowed the recordings down a whole lot. I did a weird dance
where I'd mute the already recorded one, disarm it, add a new one, arm
it, enable recording and roll. Rinse and repeat a hundred times or so.

I also used the crappy dynamic mic to record a couple of weird game
sounds as well as cheer and laugh tracks. Not sure that was smart
either. I know what you're thinking, but we are making a gameshow game
after all.

Anyway, I'll spend tomorrow morning with editing and I hope I'll find a
good way to deal with the playlist stuff. If someone knows a nice short
tutorial on Ardour workflows that would certainly be helpful.

Thanks a lot for all your suggestions.

Best regards,
Philipp