Sound Design and Dynamic Range
I’ve now mixed the audio for the
pre-title teaser and Act I of “No Children in Space” for the animatic
version. Although it sounded fine on my studio earphones, I found it
didn’t sound so good on a TV. I realized I was mixing it too “hot” —
meaning the dynamic range was too high. So I did a little research, and
now I’m mixing a more reasonable version.
Fixing Sound Levels (Dynamic Range)
The first problem was that I had not adjusted the level on the title
music, so it was pretty much filling the full digital scale (i.e. the
peaks are at 0 dBFS or “decibels from the full digital scale”), just
like it would on an Audio CD. But this is not the norm for television.
Normally, TV audio is mixed somewhat lower.
For television, you
want a narrower dynamic range (which is the difference between the
loudest and the quietest sounds on the soundtrack, expressed in
decibels) than you do for theatrical movies. That’s because people
usually don’t listen to television in an extremely quiet room. A typical
living room is much louder than a typical theater auditorium. So very
quiet sounds will be lost. On the high end, very loud sounds may be
disruptive and in any case, they can cause distortion on the smaller
amplifiers used with home televisions.
Viewing on the computer has similar (and possibly worse) limitations.
had been simply mixing “by ear” using my studio monitor earphones and
computer sound card. They’re pretty good equipment and they isolate the
environment pretty well. I found a nifty website tool for
estimating frequency response
of your soundcard + earphones + ears combination. It shows me that my
hearing is still pretty good and that the soundcard and earphones work
over a wide frequency range, though there may be some problems on the
very high-frequency end. The down side of this, though, is that stuff
that sounds good over my earphones may not sound so good on a TV set.
I found that in fact, I had been mixing with a dynamic range of about
70 dB which is probably okay for theatrical use, but is way too hot for
home-listening on a TV. I also had very inconsistent levels from scene
to scene. But what levels should I use?
I did some research
online, but although there are plenty of discussions of loudness levels
on online forums, I found the information more confusing than useful. So
instead I decided to learn by example — I loaded part of the
soundtrack from a TV series I like (“Buffy the Vampire Slayer” as it
happens), and took a look at what levels they used. I didn’t use a
formal measurement process, but just estimated by looking at the
“Waveform (dB)” scale in Audacity. Here’s what I saw:
|Type of Sound||Min||Average||Max|
|Title/Hero Track Music||-12||
|Dialog and Effects||-18||
The absolute peaks were at -8 dBFS.
This seemed like a pretty good starting point, although I didn’t see
much point in wasting the top 8 dB of scale space (thus throwing away
some resolution). So I decided to adjust this scale up to use more of
the digital scale. I’m currently working on adjusting my levels to match
|Type of Sound||Target Level|
|Absolute Peaks||-0.1 dBFS|
|Loudest Sounds||-6 dBFS|
|Title/Hero Track Music||-10 dBFS|
|Dialog, Effects, Music-only||-16 dBFS|
|Background Music||-24 dBFS|
So far, I’ve only applied this adjustment to a couple of scenes.
Mixing Progress for “No Children in Space” Animatic
As for the sound mixing in general, I have finished the before-title
teaser and all of “Act I” which is about 30 minutes in. This was about
halfway through the episode as originally conceived, although with all
of the edits, I’m not so sure anymore. The episode has probably grown a
bit in length from the original 55 minute plan.
I hope to get through Act II sometime this week, and finish up the
entire animatic. I will then be going through and improving some of the
dodgier artwork to make what’s going on a little clearer to people other
than myself. I also have to make sure that I’ve met all of the
attribution requirements and haven’t accidentally used non-free images
in my mock-up shots (some of them are photo-collages rather than