Randy Thom Interview Question.

Question

http://soundworkscollection.com/howtotrainyourdragonpanel

I watched the great video on sound works collection that was on the sound of "How to Train your Dragon". I have two questions about some of the things I heard durring the interveiw.

The first is that he said that one of his huge trade secrets in sound design are sounds that continually change in pitch just slightly. This way, they clash less with music and dialog and can be heard better through the mix because they're constantly changing. This makes a tone of sense and is genious. The question is...what would you use to do that? Sometimes I just record and create sounds without using a sampler (which would have a pitch bend), loading all those sound into a sampler would be a time consuming pain. I thought a doppler effect might work, but that's more of a sweeping EQ right?

So what would YOU use to do this?

---

2nd question :

Also in the interview they give you a few break downs of sceens with Just dialog, just music and just SFX solo'd out. They mention some of the animals they recorded to get the sounds of dragons, like pigs, horses, dogs, etc. What I noticed when the SFX were solo'd out is how CLEAN AND CLEAR the dragon sounds were!! How can you possibly record all those animals with NO outside noises, wind, birds, cars, etc?!

Is noise reduction used WAY MORE HEAVILY in sound design than I thought? Or is it just that they're on a huge budget and can record in the best conditions?

THANKS GUYS!!! I REALLY APPRECIATE THE HELP!!

Answer 1

To reflect on your first question, Randy is referring to a psychoacoustic principle that falls under something called "scene analysis." I don't know that he's necessarily read a ton on psychoacoustics...but whether or not he has, the different effects are something that you begin to pick up on after a certain amount of experience. Basically, what's going on is that sounds that have changes which occur evenly across their spectrum are picked out by the brain from sounds that do not have those changes. There are a number of ways to do this. Pitch shifting is one (Serrato Pitch 'n Time or Waves Sound Shifter are examples of plug-ins you could use), and modulation is another (either frequency or amplitude, and as slow a rate as 1 or 2 Hz can give the effect). There are a few others, but they would take a lot of time to explain here. These can even overcome the masking effect in some instances, which can make them very powerful tools come mix time.

As to your second question, it's unlikely the type of noise-reduction you're envisioning is being employed. These are big budget pictures you're talking about, and the recording of clean sounds in a controlled environment is either budgeted for or has been done prior to the film entirely. Location, time of day, mic selection and placement all contribute to capturing clean sounds that mitigate the need for noise reduction. Not to mention, the effect library at Skywalker is MASSIVE. There used to be, and maybe still is, one person dedicated solely to digitizing, and adding to the database, Ben Burtt's old analog recordings...that archiving process was going on for YEARS...and probably still is.

Answer 2

Pitch change in sounds was Randy & Phil Bensons mantra back when they came to NZ to work on the Frighteners back in the 90s and from what I observed & heard of their material, much of it was achieved in the performance of props when recording, and some was achieved using doppler. (A doppler isn't a sweeping EQ - its a very specific kind of pitch change, read up wikipedia or something) But it is also achieved in editing and through the selection and layering of sounds...

re the clean sound effects question, of course they have access to clean recordings but some of it would also be achieved via when and how various sounds are heard - edited and layered, and effectively envelope shaped via volume automation etc so that noise is masked by the louder parts of other layers...

Answer 3

The first is that he said that one of his huge trade secrets in sound design are sounds that continually change in pitch just slightly. This way, they clash less with music and dialog and can be heard better through the mix because they're constantly changing. This makes a tone of sense and is genious. The question is...what would you use to do that?

I would do it right in the recording or sample selection. So if you record something, while creating different variations, also focus on getting different pitches or timbres.

Artificially: offline pitch shifting, with envelopes? The usual one that at least every audio editor has. For more automated processing you could look at wow & flutter type effects, basically pitch shifters that are controlled by a LFO of some kind. They emulate pitch variation of tape recorders. However, real-time digital pitch-shifting is generally lower quality compared to offline rendered. I guess even a pitch correction plug-in could work. However, generally the point can't be random pitch variation, but choices made in relation to the mix or sound sequence: if you need more space to a small frequency range, you can simply pitch stuff differently so they don't clash as much to get more clarity without resorting to equalizing them. Also having slight variation in pitch or equalization in anything always sounds more lively and thus more clear or distinctive compared to a static sound. Changes in pitch and timbre work just like dynamics, sort of.

Is noise reduction used WAY MORE HEAVILY in sound design than I thought? Or is it just that they're on a huge budget and can record in the best conditions?

Noise reduction is an art of compromise. I think that first of all you should avoid ever having to use it. Secondly, I think you should only use it in relation to the mix or in a way that definitely doesn't worsen the sound. Only reduce noise in a way that is just enough, but not more. I think noise reduction is a standard practice in some cases, but ultimately you should use your own judgement about using it. Don't forget that there are also other ways to take the noise under control so it doesn't sound too bad, single-/multiband compression and narrow EQ cuts and even automation.

It's nothing specific to "sound design" though.