field recording

11 August 2004, 14:56 genre:    by mickael

In the last vr posts I have tried to provide a sound experience along the visual immersion, not the usual post-production cd track layered in the vr, but real sound samples taken on location at shooting time.

I received a few questions about this so here goes about what I use and have learned so far. I’m no sound engineer though, this is just my amateurish experiences in field recording.

The gear, the budget

My initial budget was tight so I rapidly set myself to use a mini-disc recorder (borrowed from my friend Ben $0) and low cost binaural microphones from core sound $100 delivery included.
Binaural microphones capture a broad sound environment, placed at ear level, they will record just what you hear. Think of it as the equivalent in sound of panoramic photography. The accuracy of 2 of these mics attached to eye glasses is truly amazing, capturing sound at 360 degrees.
The sounds on this site are heavily compressed and don’t do justice to the originals especially when heard through headphones as binaural sounds should.

The gear, setting it up

Mini-disc recorders are unobtrusive, you can keep them in a pocket, the mics are small and when attached to your glasses or shirt collar, nobody will ever notice you are recording. That’s what I had read before, and it is true just as long as you don’t carry along a mono-pod with a vr head and a bulky camera strapped around you neck !

n1 lesson learned: using both sound and vr simultaneously is nearly impossible for a normal person alone, if you do manage it you will most probably get the sound of the vr shooting (call that also tension breathing, clicking, etc) in the sample. If you’re not alone when you shoot let the other person do the sound. If you are alone take a backup sample at ease when you are done shooting.

Using it

After 2-3 days practice I think I know how to handle it: I will walk down to the Leisure World in Bridlington to shoot some vr and sample the great sounds of carousels and noisy attractions. Results proved me wrong.

n2 lesson learned: not just because it is noisy and rich in sound color will it make for a good sample. I returned with a brouhaha of undefined sources. Two main reasons for this, the first one being I was walking around some of the time and you need stand still for binaurals to render, even more if you want to use the results in a pano. The second reason is I had no focus in the sound: everything was confused, there was no convergence or center of attraction for the hearing.

n3 lesson learned: when sampling the sound of street scenes you need an anchor point in the sound environment to avoid the hubbub effect, choose the sound source you will want main be it a conversation, music from a window, rolling water from a tab, … and let the other sounds revolve around that. Don’t move while recording, spinning your microphones around the 360 of the scene doesn’t render well with binaural sound in a vr. Instead choose your focus and record it without moving, it will feel much more realistic.

I’m still working on, the compression rates trying to keep one minute of sound under 300kb for fullscreen and 150kb for the small pictures. As you can hear on the relevant pages I’m not quite there yet, but with the help of iTunes and Quicktime I will eventually get there (you don’t need much more, these 2 hold together the best MP3, MP4 audio encoders).

To conclude

Don’t be put off, it’s easier then it seems, and the sound is much-much-better than what a camera can record (usually just a voice mic around the 11kHz band).
It will add up to the post processing time of your vr stitching especially if you sample long sequences that you will have to edit in the end to keep the weight reasonable. But it does add a whole new perspective to vr.