Wind recorded with contact microphones

[xoadc]

Hi there,
I'm looking for some help with wind reduction.
I'm working on a project where I made a whole lot of recordings using home made contact microphones placed on radio towers, guidewires, and antennae.

The resulting sounds are beautiful and quite abstract... but there are several places where wind was captured by the contact mics (these radio towers are out on a marsh at sea level where winds often got up to 90kph).

Luckily I don't need to worry about dialogue... there are no voices, and no instruments. It's all entirely abstract ambient sound.

Still, I would like to remove the wind... but am having trouble, I've tried EQ and high pass filters, but it seems to be present across the spectrum. I'm at the point where I feel like I need to choose between leaving it in (as a textural and rhythmic element), or cutting it out and working on crossfades to cover the cuts... I'm open to any of your artistic opinions on that as well.

Does anyone have any tips or advice?
something I haven't tried or thought of?

in case you're curious, here is a link to a clip of the wind sound:

https://soundcloud.com/amanda-dawn-christie/ctower-windclip-unprocessed

It kicks in as a flutter between 7-9 seconds...
the contact mic clips i loaded to soundcloud are completely raw and unprocessed... just threw them up there for some advice and feedback.

Thanks so much for your time, and I'm really looking forward to hearing any tips or suggestions.

thanks!

xoadc

[GregV]

I don't know how well it will work in this type of application, but do you have X-Noise by Waves? It can "learn" the sound of the noise you want to get rid of and filter it out.
I've used it before to reduce string squeak on acoustic guitars, and 60 cycle hum from guitar amps.
It MIGHT work on your wind noise...
Good luck!

[xoadc]

Thanks Greg!
I don't have waves yet, but I'll try to download the free trial tomorrow to check it out.

I'm using protools 9.0

I used the EQ 3 7 band plug in to try isolating the wind. But it seems to be spread out across the spectrum...

I tried a High Pass filter with a 24db curve, and started at 60 Hz and slowly slid up the spectrum, and the wind was still present until 3.36 KHz, but then most of the rest of the sound was gone too!

I did the opposite with a low pass filter... and the wind was pesent all the way along...

I used the controls at the bottom and set the Q to 10 so I could narrow in more precisely on specific frequencies, and slowly slid along, and it was always there...

the closest i got, was adding high pass, and low pass, and pulling down a bit at the 1.5Khz mark... but then the rest of the sound just muddied.

I downloaded a program called Sound Visualizer, and in that, I was able to see the wind clearly as red and yellow streaks on blue... (but sadly I don't know what that means... I'm used to dealing with wave forms and numbers).

it's after midnight here, and i got frustrated, so i deviated and decided to try enhancing the wind instead of removing it... here is just the wind from that same clip... with high pass and low pass, and a few frequencies enhanced.... just some play... i'll sleep on it and download waves tomorrow and try to remove it.

https://soundcloud.com/amanda-dawn-c...ip-eqprocessed

[Craig F]

Take a look at RX, Spectral Repair is cool https://www.izotope.com/en/products/audio-repair/rx/

[dr_daw]

Have you tried making a copy of the isolated 'noise' that you don't like, putting it on a separate track and inverting the phase? That should cancel out the noise, but may also cancel out some of the desired sound. The problem with this noise is clearly that it has a wide spectrum, but is likely at resonant frequencies to the wind speed.

Perhaps in the future you can devise a form of windsock or windblock when recording?

[xoadc]

Thanks Craig!
I took a look at RX, Spectral Repair but it's out of my price range for now.

Instead I downloaded this open source spectral program -http://www.sonicvisualiser.org/

I was able to finally see the wind... everything I like is blue,
but the wind is yellow and red
(settings:linear scale, window 4096, and Bins = frequencies, log... colour = sunset) But I'm not sure what that means or how to access or affect those areas.

I've never done spectral analysis before. Any tips on how to make use of a spectral visualizer?

[xoadc]

Thanks Dr. Daw!

I did make a copy of just the wind... instead of inverting it, i just played with it like a rhythmic loop for fun... but I like your idea of the phase cancellation!

Since I have the copy, I'll give that a go.

As for wind protection, I did get that figured out for my later recordings. Built some custom furrys and modified my clamps. I also later invested in piezo film instead of piezo disks so I could wrap "around" the guide wires and round surfaces instead of resting "on" them.
Sadly it was trial and error so it took some time, so there is wind in my earlier recordings (the first 4 out of 13 towers).

There were 13 radio towers (over 400 feet tall) in a marsh where the winds were often upwards of 40km/hour and sometimes as high as 90kph. The towers were torn down last month, so I can't redo the recordings... this was meant to be a historical document of them before the demolition (it was one of the largest international shortwave sites in the world - gone forever now).

I'll try out the phase cancellation!

A few of my friends have suggested that I leave the wind in, because it is supposed to be a historical document and it's a windy area... It's not clipping or peaking or distorting. I'm just irked because it's not the "pure" recording I wanted... but it is what was happening on those days...

I worry that I'm just being lazy if I leave the wind in there... some people told me they like it... but ... I'm still on the fence.

[shtik]

At least for that project you can still use the 14 days demo of the Izotope Spectral Repair and effectively get rid of that wind noise before it expires..