5

The voice track in Dark, especially in the heavy rain scenes of which there are quite a few, seems at odds with what we see on screen. The words don't sound like they are being spoken in the same environment.

Someone will be shouting outside in the rain and their voice will sound like they are shouting indoors, but it doesn't really seem like the voices were recorded afterwards in a studio either. To me it seems like they were recorded at the same time as the image, but somehow lost their connection to the environment.

I was hoping someone with sound capture experience could tell me what it is I am hearing.

How do they handle voice track in such a noisy on-screen environment?

1
  • 2
    I've edited in an attempt to pull it back into 'on-topic', removing some of the emotive direction it took. I see a decent QA in here somewhere. It's possible this is a decode issue & still off-topic, but there is always the chance it's a misinterpretation of what must be faked up to achieve pseudo-reality in a loud background noise context.
    – Tetsujin
    Sep 3 '20 at 16:07
8

Personally, I thought they did a lovely job on the sound in Dark.
I mean, noticeably good, not just "It was OK, I never really paid attention"

Bearing in mind that those scenes in the rain would have to be done in post-production by looping [ADR, 'automated' dialog replacement] - rain bowsers[1] are far too noisy to get clean voice track from, and also that the rain and thunder fx are identical in the original DE and US dub.

This, apart from any other consideration, gives them full control over the ambience as played to the audience in the final soundtrack. They have totally clean dialog centre channel and everything else is in the surround channels. Looping/ADR has been sufficiently technologically advanced since the late 90s that you will absolutely never see the lip-sync slip [in the same language, of course, not the dub.]

They often use a long slap-back[2] outdoors, which though 'artificial' gives a sense of space you could likely hear if it were quiet. If they'd done the sound "as real" then it would be totally flat, no reverberant space at all, as the rain would bury any ambience completely - but that wouldn't sound convincing. [Human perception is an odd thing;)

Interestingly, the US dub does use that slap-back to a lesser extent than the DE original; hinting that two separate engineers/teams did the final mix for each language. [The EQ also seems thinner, but I didn't hear enough to tell whether that's consistent throughout.]
When I watched this right through I only listened to the original German, with subs; I don't like dubbed audio, I find it off-putting, so I've only now heard the US dub for the first time and just a few minutes of that.

If that's really not what you're hearing, or our explanations somehow diverge in other than simple 'opinion', I'd double-check your source material or decoding method, but that's off topic for Movies & TV.

[1]Real rain actually barely shows up on camera, even if it's quite heavy rain. To film 'rain' they have to use large water containers [bowsers] & pumps up to huge "shower heads" above the scene. The quantity of water that comes from these is quite staggering if you have to be under one in a scene. Sometimes, they can actually place the actors just behind the wall of water so they can stay a bit drier, but not always. The sheer deluge these things produce is not only wetter than standing under a waterfall, but is just about as noisy too.

[2]Slap-back is a term for any echo or reverberation with longer than perhaps a half second between the original sound and the first echo. This would be heard in real life if a reflective surface was 100m or so from your sound source. You can test this [if you can still find one these days] by shouting at a large block of flats from 100m or so away, or by clapping loudly [in an otherwise quiet environment, of course.]
In cinema it's used to convey a sense of space - nothing big and reflective is very close - think of the trope of shouting in mountains vs shouting in a stairwell. In Dark they tend to use it in the woods to give a sense that there is nothing nearby except hills and trees. In reality, the trees would probably kill any long refections - but there's something very 'odd' and un-cinematic if all you can hear is the source voice and absolutely nothing to follow it. It can be more disturbing than a little added long ambience. It's a trick but it's a trick that usually works to give that sense of space.

There's another trick used by sound engineers on set, which becomes very useful in post, when the separate elements are being assembled - some from live on set or location, some from looping.
There is a software reverberation technique known as convolution, which, without going into too much detail, gives the engineer the ability to add to a studio dialogue track the exact same reverberant space as they had in the real place.
The sophistication of this is such that the ambiance/reverb/echo can be calculated from just the sound of the clapper-board as it's clapped together at the beginning of a take.
This is a life-changer for sound recording.
Using this technique, there is every possibility that the ambience you hear superimposed in the rain scenes is identical to what it would have been like when it was quiet.

3
  • 1
    "Slap-back is a term for any echo or reverberation with longer than perhaps a half second between the original sound and the first echo. This would be heard in real life if a reflective surface was 100m or so from your sound source." – As a former live music sound engineer, the number 343,2 m/s is burnt so deep into my brain that my subconscious did the calculations without me even wanting to, so I might just as well post them: half a second at standard conditions (dry, 20°C) is just above 170 m, 100 m is just above 290 ms. Sep 8 '20 at 12:05
  • 1
    And yes, convolution reverb using impulse responses is amazing. Note: it's actually not that the impulse response can be computed only using the sound of the clapper, but rather that you need a short, sharp impulse, because you are interested in how the room responds to the original sound and do not want that overlaid with the original sound. So, a longer sound would actually not work, and the best is a short, sharp burst of acoustic energy from a point source. Sep 8 '20 at 12:07
  • 1
    @JörgWMittag - No, in fact to get a full convolution calculation, you actually need a 30s [or minimal 10s] full-frequency sine wave sweep, not a simple short sound; however Altiverb has specific algorithms that work on two of the most common clapboard types, just for this rough & ready convenience. Ref: audioease.com/altiverb/sampling.php My head doesn't have 343.2ms burned into it, because I've only ever been a studio engineer. Closest I've ever needed is an approximation of 1ft per ms which is close enough for time-aligning mics on a drum kit etc ;)
    – Tetsujin
    Sep 8 '20 at 13:31

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .