Field recording is fun but also challenging. A while back I posted a comparison of microphones and stereo techniques. I thought I would add some details about the day I recorded that material which was on new years eve when I decided to give it a go to record some isolated fireworks going off. I purchased an assortment of about $400 worth of consumer rockets and shells.
Scouting Location The day before recording I needed to find a location. Setting off fireworks in Sweden is restricted to new years eve unless you have specific permissions so I wanted to take advantage of that and record it when most people were expecting fireworks to go off. I had purchased a fair amount of fireworks that would take some time to set off, I expected it would take a couple of hours or so I wanted to find a place that was far away enough from residential areas but still accessible by car.
I knew of a location a kilometer or so away from the main road, pretty far away from any residential housing. It's near a motocross track where I ride in the summer, there is also a race track for "folkrace", and a little strip where people fly remote controlled airplanes.
There's a mud road leading into this place and I remembered scouting that location a year or so earlier. Back then it was late autumn/early winter, it was snowing and it was nearly midnight. That time I had a sedan car and my wife and three kids were in the car and we got stuck =) My wife had to push while I reversed, but that's a different story.
Anyhow, this time I was there in a 4x4 car, it was during the day, and it was not snowing. Still, I hit a weak spot of ice cover and wheels sunk into the mud, but this time it was no problem reversing out of there.
I had found my location. Now it was time to prepare for the next day.
Preparations The evening prior to recording I gathered the equipment and charged all the batteries. I decided to bring the following equipment:
Recorders and Mixers
Sound Devices 788T SSD
Sound Devices 633
Sound Devices 442 (extending the 633 to record 6 channels 192 kHz instead of 3 channels)
Pelican case to pack the microphones in. Look, it's a box full of dead kittens!
$400 worth of Consumer grade fireworks
A blanket to put equipment on
2 storm lighters to set off fireworks with
Bin bags for litter
Paper and pen to record mic setup
A friend to help set off the fireworks
I should learn that prepping always takes longer than expected. I started about 8pm, and I think I went to bed at 6am.
I picked up a friend of mine who would help me out to light the fireworks as I focused on the equipment. We unpacked everything from the car and it took about an hour to rig it all.
So, I had 10 microphones to record with and 4 of them (the Sennehisers and the Schoeps mics) were connected to the SD 788T and the rest to the SD 633, 3 of which had to pass through the SD 442 as the 633 only has 3 mic inputs (and 3 line inputs.)
I spaced out the microphones to get some different perspectives:
Conditions It was a beautiful day but a bit of a breeze so wind protection was essential. It was also fairly cold, but I am fairly warm blooded and I tend to keep my temperature very well, especially when I'm occupied with keeping everything under control for the recording session so that was not a problem for me. My friend, on the other hand, was freezing before the recording session even started =)
Recording Begins But then it didn't! Daylight is fading fast in Sweden at this time of year so I was in quite a rush. As the final mic was rigged, I joked with my friend saying "did you bring any lighter?" We both laughed as I took out two brand new lighters that I purchased the day before. Our laughs quickly turned to frustration. Both new lighters had no lighter fluid in them. Don't take ANYTHING for granted =)
I rushed off leaving $30'000 worth of equipment and my friend in the remote location driving off to the nearest petrol station to buy new lighters. I bought 5 of them before hurrying back. 20 minutes of precious time was lost in an already tight schedule.
Finally we got started. My friend burned his finger setting off the second rocket so he was out of action. Pictures were taken for all the fireworks that was set off to document the sounds. I ran back and forth setting of rockets and shells returning to the recorder location to make sure levels were OK not to clip any audio data.
Setting off fireworks, recording 10 channels of audio into 3 recorders/mixers, ensuring appropriate gain and levels are set for optimum audio, making sure all settings for bit depth / recording frequency / and limiters is configured correctly, documenting everything that is going on, is quite a challenging and stressful thing to do.
You don't want to come home only to find out that quality settings were wrong, tracks were muted, or all the audio is clipping or recorded too low. Not when investing time and money to record in the single day of the year when it is legal to do so.
It all went well and I was happy with all the audio that was captured. I put together this video that records one out of many fireworks that were recorded that day:
The video speaks pretty much for itself in terms of how everything was set up, the settings I used, and how the recording session went down. It compares what the different microphones in their different perspectives and with their different characteristics captured. The video also contains a section where the audio is played back at greatly reduced speed. I use fireworks recordings to create all sorts of explosions so pitching down the sound while maintaining high quality is very important to me. Hence recording at 192 kHz which in most other applications would be overkill and unnecessary unless you plan to warp the audio the way I do.
What went well? Over all I was thrilled with what was captured. 8 out of 10 channels had what I consider to be very good levels. Most of the fireworks that were set off were captured cleanly and I got hundreds of excellent launches, rocket explosions, and shell explosions. Over all the equipment was well organized and I had planned mic distances, recorder settings, and channel inputs in advance making it possible for everything to run as smoothly as it did.
Important Lessons Learned Every time I do field recording I learn something new and this time was no different. Some of my key takeaways from this particular recording session:
Plan more time for prepping cables, recorder equipment, etc. Do it at least 2 days beforehand and don't leave it to the last night.
Test everything, including and maybe especially new stuff that may not be audio related but oh so important. In my case the two lighters that were essential do set off the firework that did not work despite being brand new.
Give myself more time by starting earlier to set up. It got unnecessarily hectic considering I had to pick up my friend who lives 40 minutes away and then return, etc.
Don't set up recorder location among the microphones. Since I had to light the rockets and run back I had to run back right to where the microphones were so some launches were spoiled by footsteps. I kept breathing under control, however =) If my friend would have been able to set off the rockets and step to the side in another direction, this would not have been a problem - but having to set it off myself it became a much bigger problem.
Don't leave plastic bags by the mics. The wind caught them at times which generated noise.
Prep exactly which items to launch in what order and prep with printed sheets to tick off and write recorder file as you fire off each item.
Warn friends even more than I did how long something like this takes. And how cold it can be.
Once we packed everything up it had gotten dark and we left the location. I drove my friend back home to his then I rushed back to my family to celebrate new years eve together. I could not help myself so I also rigged the 4 microphones on our roof while I set off fireworks with the kids in the garden and I kept the microphones running to capture some nice midnight fireworks going off in my neighborhood.
I have become somewhat comfortable with different stereo techniques even though I still have very much to learn and even more to master. I took the opportunity to purchase a variety of fireworks and rigged an array of microphones in a somewhat remote location while it was still daylight on new years eve. In total I brought 10 microphones that I placed in different stereo and mono configurations ranging from a fairly close distance of 25 meters, to a remote distance of 100 meters.
It will likely come a time when I write a full post about the fireworks field recording session but for now I'd like to focus on two pair of microphones in particular; a pair of Sennheiser MKH 8040 cardioid microphones arranged in the ORTF (Office de Radiodiffusion Télévision Française) stereo configuration, and a mid-side (M/S) setup consisting of a Schoeps CCM41 supercardioid and a Schoeps CCM8 figure 8 microphone.
Most of my stereo recordings so far have been ORTF recordings which is probably the simplest configuration to grasp. Basically, it is two identical cardioid microphones that are spaced 17 cm apart and angled 110 degrees from one another, which sort of mimics the way human ears work. The upsides to this configuration is simplicity and a nice realistic stereo width to the sound. The downsides, on the other hand, are that 1) there is no microphone pointing straight ahead so a good mono recording in center field will be missing, 2) there may be phasing issues when the stereo recording is mixed down to mono (i.e. the left and right channels may be cancelling each other out to some extend since the microphones are spaced apart) and 3) the stereo width is final, you can't really make it narrower or wider by post processing.
Despite its downsides, I like ORTF because it sounds great for stereo ambiences which is mainly what I use it for. In such cases I rarely require a mono channel facing forward (e.g. forest, beach, city, public places) and to avoid phasing issues all together, for a good mono sound, the simple solution is to just use the left or right channel single as mono.
[caption id="attachment_83" align="alignnone" width="840"] My pair of Sennheiser MKH 8040 in Rycote ORTF without the blimp cover[/caption]
There are many cases, however, where the idea of capturing mono compatible and directional audio would be ideal. The Mid - Side (M/S) stereo technique allows this and as an added bonus, you can even decide how wide you want the stereo to be after the sound was recorded. Throw the idea of how human ears work aside for a moment.
The M/S technique uses two different types of microphones. A front facing "mid" microphone (cardioid, omni, or in my case, a supercardioid) captures whatever sound it is aimed at in mono. A second figure 8 "side" microphone located directly above or under the mono microphone captures sounds to the left and the right.
The M/S technique scared me for a while since the captured audio isn't compatible to play back without first being processed. The raw M/S audio is "encoded" which means that the raw audio captured from M/S contains the center audio in the left channel and the side left/right audio in the right channel. This material won't play back correctly without first decoding the audio to standard left/right stereo.
There are different tools and methods to perform the decoding. At a basic level, the mid channel (usually the left channel of the encoded audio) is treated as mono resulting in being identical in the left and right channel of the final decoded stereo audio file. The side channel (usually the right channel of the encoded audio) is treated by being converted into stereo audio where the right channel is phase inverted. Once the side channel has been prepared it can be mixed with the mid channel and depending on the amount of mixing you can make the resulting stereo file anywhere from very narrow to very wide.
[caption id="attachment_80" align="alignnone" width="840"] My Schoeps CCM41 + CCM8 M/S in Rycote ORTF without the blimp cover[/caption]
Once you have performed the M/S decoding a few times you'll learn that there is no rocket science or magic going on. The great flexibility of M/S does, however, come with some downsides. Decoded M/S files can't be used as any old recording without first being processed, and if sold as part of a library, you have to be extra clear to point out that the audio is M/S encoded and what the consequences of that are. A way around this is if you first decode the audio, maybe into a narrow, and wide version, for distribution or inclusion in an sound effects library. You still have the great advantage of selecting the stereo width after the audio was recorded.
Now that the basic differences between ORTF and M/S have been outlined, I wanted to take this opportunity to share a comparison between the two:
The audio example above contains first the ORTF recording, followed by the M/S recording decoded with 50/50 with side channel for a wide image, and the final section is a mix between the two.
During my fireworks field recording session, I had my Sennheiser MKH 8040 ORTF setup and Schoeps CCM41+CCM8 M/S setup, located in the same place, both aiming towards the fireworks with a slight angle up into the air.
I recorded the 4 audio channels into the same Sound Devices 788T recorder at 192 kbps/24 bits. I set the gain conservatively not to clip the peaks and the limiter was disabled. This was a deliberate choice because I didn't want the limiters to alter any audio and I was hoping there would be enough quality and room to raise the gain with a faster post limiter to remove the extreme peak audio without having a sweeping effect on the following echo tail.
In post, I first normalized the audio which increased the gain by about 5dB. I then ran a maximizer plugin increasing the volume by another 35dB while limiting the peaks. This is a substantial increase in dB and the noise floor was raised in the process.
I was quite surprised at the difference of the two recordings. The Sennheiser pair picks up much deeper bass frequencies and the Schoeps setup is much "snappier" and sharp but lacks those deep frequencies of the Sennheiser mics. I also feel that the ORTF recording lacks a bit in the center and the width of the M/S recording feels nice. Mixing the two together is the closest representation of what I recall actually hearing on site.
This year I went to the Nynäs Offshore Race which takes place only a 30 minute walk from my home in Nynäshamn. I brought my new "field recording kit" that may not be the most portable setup in the world, but it gives me the best option for recording directional spot mono audio using a Rode NTG3 shotgun microhpone and environmental ambiences and stereo effects with a pair of Sennheiser MKH 8040 micorphones. The mics are housed in wind-protecting Rode and Rycote blimps with dead kittens (those huge furry things to block most of the wind.) I record the 3 channels of audio into a Sound Devices 633 field mixer at 192 kHz 24 bit audio.
I recorded quite a few powerboats as the passed by - but getting usable game audio is quite tricky. I'll tell you why:
First of all, there is a PA system with an announcer that feels obligated to speak non-stop which is great for the audience but terrible for someone who wants to capture the boats and nothing else. I had to find a spot close enough to the boats, but far away enough from the diesel-powered speakers pushing out a distorted commentator voic.
Then there is the audience, they tend to chat and cheer a lot so I had to find a spot far enough away from most of them - luckily the powerboats are loud so the gain is not set too high which reduces the chance of spectator voices.
Another challenge is that you don't really get that close to the boats, they have to drive where it's deep and safe enough which isn't always that close to land.
The splashing water is another parameter to consider, but again, not loud enough to be too great of a problem.
The biggest problem, however, is that the boats pass by producing a sound that fades in and out and shifts in pitch due to the doppler effect. By default they will be great if you want to have the sound of a passing powerboat - but you could never apply that sound to a boat in a game since it simply doesn't loop well.
Turning the passing powerboat sound into looping game audio
I put together a video where a single powerboat passing by is turned into a short loop of game audio that could be used on a powerboat. If you need to do something similar, or if you are just curious what work goes into converting the sound, have a look:
I ended up using the audio from the Rode NTG3 shotgun mic this time since I didn't need it to be stereo and it produced the longest sustainable sound as I aimed the microphone at the boat as it passed by. It also contained the least polluted sound since the design of the shotgun mic blocks off-axis sounds fairly well.
This sound loop, among many others, will be included in the next update of the Universal Sound FX library.