zs_overman (zs_overman) wrote in bloodspell,

Episode 8 Did You Know

Walla Walla

No, I'm not talking about the largest city in Walla Walla County. I'm talking about rhubarb. Episode 8 brought a return to a familiar locale (the pub), and introduced us to a new one (the arena), both of which are packed with noisy people.

A big part of making a convincing crowd noise is getting the right size. But another challenge is adapting a smaller sound sample to a longer scene without having a "looped" feel. The first strategy is obvious: crossfading the sound so its seams are not audible. Crowd noise samples tend to lack consistent level and texture throughout (lack of it helps them sound authentic, actually), so a longer crossfade is preferable to a short one, the length depending on the degree of variation between the level/texture at the start and end of the sample. If you can time your crossfades to occur while something else is happening in the foreground (i.e. a line of dialogue or some prominent foley), so much the better to mask your edit points.

Even taking these measures, and staggering the timing of the crossfades, it's still relatively easy to perceive some "cycles" in the sample repetition. That's where layers come in very very handy. Having multiple layers of different crossfaded samples (i.e. same basic sound, NOT the same sample), with much variance in the timing of those crossfades, one can take several 20 second samples and craft 3-4 minutes of crowd noise with no readily discernable repetition.

In the tavern, there are three different crowd noise samples going on, and in the arena, there are five woven together, six in some spots. One of these arena samples is from a brief session Strange Company recorded in person (if I'm remembering right, from a Thom Tuck show in Edinburgh??)... you can hear some of that session, including the coaching of the crowd, as the audio underneath the Episode 8 credits. Sounded like a fun time for all, and those richly textured foreground crowd noises are really what made the whole thing work. You'll hear some more crowd action in Episode 9 on Wednesday the 20th of September.

What? Huh?

An additional challenge of working with noisy crowd ambience is akin to the trouble one would have when actually in such a place: being heard. In the first Eschaton movie, I seem to recall a club scene where the dialogue was allowed to be swallowed by the ambient noise, and subtitles were used to keep the audience connected. First time I saw this technique was from David Lynch in Twin Peaks: FWWM; not sure if that was Hugh's inspiration or not, but it was an interesting device.

In any case, that wasn't on the docket for Jered and Carrie's conversations in the arena. The ideal goal in such a scenario is to preserve as much of the levels and intensity of the crowd noise as possible, while still hearing the dialogue without trouble. Music producers are confronted with a similar challenge when faced with a song mix containing two instruments of similar timbre which should each be easy to discern in the mix. The instruments tend to compete, and the result is muddiness where distinction is desired. Some of this can be addressed by panning, but a more effective technique (and one more applicable to our situation) involves equalization (EQ). By emphasizing a key frequency range of one instrument, and proportionally de-emphasizing that same range in another, the instruments stop competing quite so fiercely and tend to more peacefully coexist.

This was the approach which aided Jered and Carrie being heard over the arena noise. Yes, there was level reduction in the crowd ambience, this was crucial. But much of the blend was achieved by giving a slight cut to the 250-350Hz and <>2000Hz ranges in the crowd tracks. This created a nice pocket for the dialogue to fit into without as much competition, and a subtle boost to those same ranges in the dialogue tracks for that scene brought them to the foreground even more.

Keep in mind, the frequency ranges on which one should focus are entirely situational to the sounds you are trying to "blend with distinction", so don't take those numbers as any kind of hard and fast rule. Examine the dominant frequencies in the two sounds you wish to play friendly, and the ranges where they are both strong are the best candidates for this kind of adjustment. Finally, adjustments to EQ should generally be subtle, not drastic, unless your intent is to greatly distort the sound (e.g. for a special effect).

  • Welcome to BloodSpell.com

    Hello, and welcome. The BloodSpell project is finished now - over just under five years, we created a hit Web animated series, turned it into a…

  • The Last BloodSpell Production Post Ever

    Four years and 10 months ago, a drunk French guy told me that Strange Company, the production company I run, and the company that made BloodSpell,…

  • BloodSpell: What Next? Making of

    Yes, it's the last release ever on this site, barring miracles. In this documentary, the cast and crew of BloodSpell, including me, discuss how we…

  • Post a new comment


    default userpic

    Your reply will be screened

    When you submit the form an invisible reCAPTCHA check will be performed.
    You must follow the Privacy Policy and Google Terms of use.