Sound Design Stack Exchange is a question and answer site for sound engineers, producers, editors, and enthusiasts. It only takes a minute to sign up.
Sign up to join this community
Hey all,
I'm working on a sci-fi actiony flick right now, very low budget so mixing in surround is not an option. I have a scene where a robot with a jetpack is landing behind someone, off screen in the distance. I've been playing around with EQ and verb and cant get a realistic sound. Basically what I'm asking is; what is the best way to make things sound like they're coming from behind you using stereo?
The only way I've been able to do this sort of thing is by fooling around with phase reversal and cancellation.
You know when something is out of phase in headphones and it sounds like it's coming from behind your head?
I tried to do the same thing you're doing on speakers and I spent a little time on it with no real success. But you might have brighter ideas than me when you try it out.
I would suggest rolling off all the highs and release it once it's on screen.
Or make a REALLY wide stereo spread of the jetpack and then instantly pan them both to center once the jetpack whips onto screen.
You can also research how some audiobooks use this "2.0 Surround" technique - I know nothing about it but I've seen audiobooks with stickers on them saying "Stereo Surround Sound!". Dunno if it's legit or what.
Sorry I couldn't be of more help =/
Well one method is to invert the phase of one channel of the stereo signal. It's a trick long used in moviemaking. When the sound has phase correlation and definite stereo image ("phantom"), we can localize it by panorama on stereo speaker setup. If you rotate the phase of left or right channel, a listener can't localize it in front hemisphere of him, thus he makes a conclusion that the sound is not in front of him but behind rather. If you're having mono source, just pan it to one side and mix in a phase inverted copy of it at opposite stereo position. But still this effect is really weak on my judjement when using front stereo speakers.
If you're targeting headphones, you have much better oportunities of simulating sound position. Exactly what you need is binaural spatial position effect. WaveArts Panorama is one of them, there are also some Max (cycling74.com) devices including one for Ableton Live. The key concept is called HRTF, which is the specific spectral processing happenning when sound wave is travelling through our head (or body in general) when reaching one of ears. That's the most reliable cue to our brain, though it's not absolute by any means, but researches have shown that vast majority of people can localize digital HRTF-processed sounds as intended in most cases.
There some tricks if you're trying to put too short (<100 ms) sounds behind a listener, as well as there re some problems concerning sounds without enough high frequencies (>3k) - those can't be localized by our hearing well also.
Most of these effects are achieved through phase trickery or modulation. They tend to work well only if the sound source is moving within the stereo field (like one of the presets in Waves MondoMod). Also, they only work in the sweet spot or with headphones.
Although, it's going to sound quite odd though to have a single sound behind the listener (if you manage to get it work). More than achieving realism you are going to pull the listener/viewer out of the narrative. You might be better off rolling some HF to simulate how we hear sounds behind us.
Also, from a psychoacoustic perspective, we do find it difficult to localize sounds directly behind us - which is why we move our head around to place the sound source in front of us where it is easier to locate it using inter aural time/delay differences and also reflections off our shoulders/pinna.
Hi Kyoti
I hope I am understanding correctly in that you already have a recording of the robot with the jetpack landing, but that it does not sound realistic perceptually. If this is the case, the quickest and easiest solution would be to use a convolution reverb with the appropriate HRTF impulse. There are several factors that affect sound perception, such as time, intensity, pitch, etc that you can use to your advantage to tweak the results until you are happy although, considering your budgetary constraints, you need to decide how much time you are willing to spend. You could also try some of the following techniques, but bear in mind that the time vs results ratio is not proportional:
1) The robot is obviously moving down and most likely, I am guessing, toward you? This will affect the characteristics of the sound in relation to time. A gradual increase in pitch will provide the effect of motion toward you (Doppler). Also, higher frequencies will gradually become more evident as it gets closer to you.
2) Amplitude will also gradually increase as the robot is approaching you.
3) The recording should correctly represent the space in terms of reflections from surrounding buildings and objects. Your words "in the distance" give the impression that it is a large open space. Careful application of reverb can assist you to recreate a fair representation of this space. This can be very time consuming and it is doubtful you will improve on the convolution reverb.
4) Ideally you do not want the same sound coming from the left and right channels, as this is the information that your brain uses to create a stereo spectrum and automatically place the origin of that sound in front of you. Obviously, it will not work to use only one channel, as your brain will place the origin of the sound to that side and it will still not be behind you. It will not help to use a stereo file, as there will almost always be elements that are present in both channels. This then gives your brain sufficient information to create a stereo spectrum, which it will automatically place in front of you.
One way to fool your brain is to increase the pitch of one channel by 1-2 semitones and decrease the other by exactly the same, although I find this technique is more effective with a musical source.
Another simple trick is to use a twin mono recording, but invert the phase of one channel. The phase cancellation should prevent your brain from creating a stereo spectrum and fool it into perceiving that the sound is coming from behind you.
Personally, all things considered, I would probably take the approach of using the convolution reverb and then tweaking it to taste with 1 and 2 above using straightforward pitch, amplitude and EQ envelopes. Hope my "convoluted" reply helps.
Maybe (really just maybe)
get a binaural mic
Place your designed robot with jepack in mono in a speaker behind the mic, in a room similar to what you're seeing. Give yourself a sync pop before the sound so that you can have it in sync later on.
Record and import.
profit?
the effect would only really be in full force when wearing headphones, but even through stereo speakers the perspective should be as close as you would hope to achieve with other means. Maybe also try doing a couple of takes where you move the speaker around a bit in space as the sfx is playing back.
never tried it really, but this would be my first attempt.
There are three plugins that I have which can simulate a 2d sound coming behind you stereo field.
One is called Wave arts Panorama, another is the sampler, Kontakt, there is a effect for surrounds in that program, and Waves MondoMod I think has a back stereo field.
A short list of things to think about from your recording:
Bass correction - If the mic was close, the bass is going to be boosted. A low-shelf of down to -20db (likely less) at around ~100Hz can put the recording in real space.
Treble roll off - If you want the sound to be far away, you'll need to pull off some high end. To do it the complex way, here's a calculator. To do it the easy way check out this other question
From here, the rest of these can be tested by snapping your fingers in front of your face and behind your head:
In short, a sound from behind will hit your ears from behind, filtering it through some very aggressive and hard (impossible) to simulate EQs (your head/ears), then hit walls if they're there (filtering less through your head meat this time), then hit a front wall if it's there with the final arrival being the least meaty and least wide.
One last note: the meaty absorption of your head is different than that of the ears. So draw a line from the source to each ear and see what kind of absorption you're getting. From there, it is a measure of distance and angle that decides.
Thanks to everyone for the help. I played with the phase a little, widened the stereo image and played with the verb and got a lot closer to what I was looking for. Everyones responses really helped.
I'd post the clip for you but the producer is being a little finicky on sharing anything about the movie.
