Atlantis is a short animation created by Macauley Bell, which depicts an underwater scene from the novel 20,000 Leagues Under The Sea, written by Jules Verne. The animation doesn't depict anything specific from the scene. It is more of a snapshot of time within the world that Verne has created. I was brought on to the project to create an underwater sonic environment and also to give the scene an ominous feel.
In preparation for this task I researched recording techniques that would either allow me to record underwater or manipulate an above water recording into sounding as though it is underwater. Unfortunately I didn't have any luck with the former. I struggled to find techniques online - outside of using a hydrophone - and techniques suggested by peers weren't convenient to my situation. These suggestions included running a long tube under water with a microphone at one end. Unfortunately I didn't have a tube or access to a large enough body of water in the time allocation I had. I also had concerns that a tube would create strange resonant frequencies. Another suggestion was simply to cover a microphone with a balloon or condom and submerge it in a bucket of water. In the case that there was a leak while submerged I didn't think it worth the risk to destroy equipment or electrocute myself in the process.
This left me with the option of performing an above surface water recording, which meant finding a way to manipulate the recording to make it sound underwater. Of course, the first thing that I realised is that I had no idea as to how sound travels underwater. The basic points that I learned through research is that:
- While sound travels at around 340 metres/second, sound travels underwater at closer to 1500 metres/second.
- Temperature still affects the speed of sound underwater. So generally the speed of sound slows down more at greater depths. However, at a great enough depth (approximately 1200m), sound begins to speed up again. This is due to the immense pressure of deep oceans and that sound travels more efficiently through more solid objects.
- Due to the up-down-up-down bending of low frequency sound waves at low depth, these frequencies are able to travel thousands of metres without losing significant energy. It is also worth mentioning that these waves are significantly longer underwater (20Hz through air = 17 metres; 20Hz through water = 75 metres).
What I gathered from this is that I should be concentrating on low frequencies more than anything. I also considered that there would be a lot of reflections from the ocean floor, which may cancel out high and sub frequencies.
As this project was only a short animation on a short timeline I had intended to utilise my own equipment - C-2 microphones and a Tascam 322 Interface. Although not the most ideal equipment, I expected it to be the most convenient form of recording. Unfortunately, I had issues between my computer and interface, which I had not run into before. I tried to troubleshoot the issue but decided not to waste my time on it and opted to borrow some location recording gear - F8 recorder and MK416 microphone. I immediately realised this is what I should have done all along as the F8 and MK416 are miles better quality than my personal gear.
By this stage I had wasted any time I had allocated to finding natural water sources (creeks, etc.) that would be suitable for recording, so I opted to just record some water samples in my bathroom. I found that filling my sink and wading my hand through the water created a nice trickling sound without harsh transients, which I would expect not to be present underwater. For this I placed the 416 on axis towards the centre of the water from the edge of the sink (similar to how a mic would be placed for a snare drum). I next recorded the sink being emptied first with the mic placed next to the sink and next with the mic placed by the drain on the floor to capture the water travelling through the pipes. I also recorded some toilet flush samples as they could be manipulated in different ways to create something strange and unique.
After some experimentation with manipulating the samples I decided on using the 'splashing in the sink' sample along with a loop of the 'drain emptying' sample (the one recorded at the sink, not the drain). Originally, I also used one of the toilet flush samples in reverse and time shifted it to make it slightly longer. This was to emphasise a part of the animation where the camera quickly climbs a volcano from base to crest. I think it had a good effect apart from there being a slight gurgle at the end of the sample, which was much too obvious and had no place in an underwater environment.
The purpose of using the 'splashing in the sink' sample was to give the water movement. The camera moves around and there are clearly schools of fish rushing by so I wanted to compliment these visuals, but in a subtle way as we perceive sound as non-directional underwater. The 'emptying the drain' sample was used more as ambient noise/rumble. There is a natural rumble that we get when our ears are covered - for instance, when you cup your hands over your ears or hold a large shell up to your ear, the sensation of being underwater is similar. Although this cupped ear simulation isn't entirely accurate to how sound behaves underwater, it is how we perceive it and we wouldn't recognise it if it were any different.
To get these above surface recordings to appear as though underwater I concentrated on frequencies below 400Hz. I figured that this would eliminate enough of the high transients that we associate with water splashing at the surface and would also leave plenty of room for the guitar sample to slot in. The two water samples were EQ'd in similar fashions, however I gave the 'emptying the drain' sample a boost in the sub region an also gave it a second boost using the 'bark of the dog' plugin in order to enhance the low frequencies.
I felt that the water samples sounded too present in the mix but i knew it would be a bad idea to add reverb as sound reflections don't behave the same way underwater as they do through air. To make the water samples appear less present I simply duplicated the water sample tracks within Pro Tools and flipped the phase on the duplicated tracks. I dropped the volume of these tracks then slowly introduced the volume back in until the water had a nice placement within the mix. In retrospect, perhaps I could have used a delay with little feedback to enhance this further.
With the guitar, I wanted to create something eerie and ominous. For this I decided to use pinch harmonics. A harmonic on guitar is achieved by plucking a string while touching string lightly over the 12th fret and pinch harmonics are similar except that you press down on a fret while you perform the light touch and pluck motion one octave above where the string is being fretted. I find this is great to use when only an acoustic guitar is available as the timbre of the sound is not specific to guitar and can be achieved without an electric guitar and effects pedals. In processing the guitar track I removed all frequencies below 400Hz to give room to the water samples and I also dipped out some frequencies around 8kHz to remove the attack of the plucking of the string.
Usually I would DI the guitar when doing this technique to avoid this issue but didn't have the option on this occasion. I also added two delays to the guitar: one set at a time of 20 and 40 ms to make the guitar appear wider; the other was given a much longer delay time of 800 ms with plenty of feedback and was then routed directly into a reverb plugin. The purpose of this second delay was to give the guitar space and hopefully lose any directionality associated with it.
Overall I feel that what I produced had the desired effect that both myself and the animator intended, although it took me a few attempts to get there. The original audio that was handed over to the animator was far below the desired loudness standards but the animator seemed unfazed but this (this is also the audio currently being presented online but is scheduled to be updated shortly). I have since corrected the overall levels of the audio to meet the FreeTV Australia standard of -24LUFS. Although the intended release platform was Youtube, which normalises their audio to -13LUFS, I thought it best to still aim for -24LUFS as this will still sound fine on Youtube but will also not degrade the audio should the animator choose to release the product elsewhere. Please enjoy the animation below.
F8 Recorder [Image]. Retrieved April 3rd, from https://static.bhphotovideo.com/explora/sites/default/files/ts_65.jpg
Sennheiser MK416 [Image]. Retrieved April 3rd, from http://www.pinknoise-systems.co.uk/user/products/large/Osix_2-416.jpg
Nieukirk, J. (2015). Understanding Ocean Acoustics. Retrieved April 3rd, from http://oceanexplorer.noaa.gov/explorations/sound01/background/acoustics/acoustics.html
Collins, A. Underwater Sound Propagation. Retrieved April 3rd, from http://www.arc.id.au/UWAcoustics.html
Pro Tools 12 [Image] (2015). Avid. Retrieved April 3rd.
Macauley bell (2017, March 22). Atlantis Final Product [Video File]. Retrieved April 3rd, from https://www.youtube.com/watch?v=t3ZU5c5_y-Y&feature=youtu.be