As part of a Yoga Nidra video project, we had to make some new field recordings of typical English countryside bird song. These recordings were to be used in a mix with Tibetan gongs, singing bowls, a female voiceover and an underscore. So, making sure we had clean recordings with a tightly managed frequency range was very important, so that the bird sounds could easily sit in a broad mix and still be clearly heard.

We’ve done this type of field recording many times before but because of tight project deadlines it wasn’t possible to choose an ideal site i.e. a remote woodland well away from traffic noise and persistent over-flying commercial aircraft. So we had do the best we could; recording in a local wood not far from a residential area and the dreaded M25. Visiting the site beforehand made it obvious that we would have to do quite a bit of work on the recordings in post, but luckily, this particular place had limited footfall, especially that early in the morning. So that was at least one blessing!

To get the best ambience, I wanted to use a large diaphragm condenser stereo microphone, so I pulled out an Audio- Technica BP4025 Large Diaphragm XY Stereo Mic from the mic cabinet and paired it with a Rycote Windjammer kit WS3 and a Rode carbon fibre boom pole for support. I charged up batteries for the Sound Devices MixPre6 we use for field recording and put everything into an Orca OR-280 sound bag, which is the perfect carrier for this gear.

Next morning, just after dawn, I was off to the woods!

One of the helpful features of the AT BP4025 is its polar pattern and frequency response. I’ve found that when trying to minimise inescapable broad band traffic noise it’s helpful to carefully rotate the mic so that the low frequencies and mid band noise is minimised (behind the mic).

The frequency response of the AT BP4025 shows a kick up at around 3Khz and again at 6Khz to 12KHz which can be useful for recording bird song, but the ruler flat response below 1Khz wasn’t any help at all on this occasion so the low cut filters on the MixPre6 were used to reduce some of the incoming low frequency content.

The wood I was working in was less than five miles from the M25 and on a windless morning it’s impossible to get clean tracks without picking up heavy traffic background noise. I was running the MixPre6 inputs at around 55dB gain and carefully picking locations to minimise noise but as you’ll hear, that was quite a challenge.

I spent around two hours recording and came back to the studio with a wide range of material, but it was obvious from headphone monitoring that there was going to be substantial post processing before selected tracks could be incorporated in the planned mix.

Bird song has a lot of high frequency energy and a very wide dynamic range with surprisingly loud transients, so it’s vital to set mic gain carefully to avoid clipping while still getting adequate levels for the quite bird sound or the ambience you’re seeking. Also it’s important to keep changing position so that you get a good range of birds and their dynamics. Still, morning choruses can be very loud and dense, so I was trying to capture single bird song so that it could be featured in the mix as desired.

So, now to the post processing. You can of course do this kind of work in Audacity at no cost, or work ‘blind’ using the Accusonus tools, but I’ve learnt that being able to see and immediately hear what you are doing and to be able to use a variety of tools in the same software package means you can work more efficiently in the long run and also try different approaches without moving files from one environment to another. For this reason I chose to work in iZotope’s RX7 Audio Suite

As you’ll see, the selected bird song file shows strong content from just below 2kHz all the way to over 20Khz, whereas the mid to low frequency broad band noise from traffic below 1KHz is fairly constant at approximately the very same level as the target bird song material. The image below show the left and right channel recordings with the sound spectrum analysis in the background and the waveform overlaid on it in blue. Looking at that you can see just how much energy is recorded below 1Khz

It’s easier to see this low frequency data in the following image, made using a detailed Fast Fourier Transform spectral analysis of the overall file: It’s particularly bad at 500 Hz and below despite the use of low cut filters in the MixPre6.

This is the selected original stereo file, unprocessed. Sadly this player only allows .mp3 files at 128Kbps:

So, how to process these files in the most efficient manner and get the best possible files? Well, the previous week I had been working on female voice files and used all three tools I’ve mentioned above and I realised that sometimes simpler is better - and certainly quicker. Looking then at the spectral analysis and realising that all the target material was actually in the upper frequencies, after much experimentation and careful listening tests, I decided to use a rather steep low cut filter acting from about 1800Hz downwards and a high cut filter to take off unuseful high frequencies above about 20KHz but kicking in at around 16KHz.

You can see the dramatic effect this EQ had in the following image, which shows the very high and low to mid content almost completely removed (illustrated on the upper, left channel with the shaded boxes). You can also see the effect in the bottom left spectrogram:

After this step it was now safe to apply some mild compression and make up gain to tame the peak levels of the foreground birds in the recording and also bring up the ambient background bird song to give a little more of the contextual atmosphere. Finally I levelled up the recording by targetting an LUFS Loudness of -23dB LKFS to EBU R128 standard.

This was the final result (again unfortunately in .mp3 128Kbps format):

This file, along with all the others, also treated the same way was used to compile a composite field recording which was succesfully used in the video piece mixdown. It just goes to show that you can pull back even some otherwise unuseable recordings with a bit of careful post processing!