"What is it that the windshield does that the engineers can't reproduce in the software?"
..simple answer is, if the wind noise is recorded, of course it's embeded in your audio...

..using a windscreen, helps to prevent wind noise being embeded to begin with...

..an ounce of prevention, is worth a pound of cure!..

Windshields do not filter anything out, they stop the wind from reaching the mesh on the outside of the microphone and creating the noise in the first place.

How I think they do this is by creating a huge number of tiny turbulances as the air flows over the hairs. If wind reaches the metal or plastic outside of a mic as it flows over the mesh or slots these act as resonators, much like a 1000 whistles. That makes a lot of sound which gets into the mic and gets recorded. Once that sound is created nothing can get rid of it.

Note that wind itself is silent, even when it flows over smooth surfaces (like the wing of a plane) however place the right shaped thing in it's path and a lot of noise is created. That's how every wind instrument works, either resonating cavities (pipe organ, flute) or vibrating reeds (woodwinds) .

Bob.

..where are you guys located?...
.. I've always heard them refered to as windscreens..

..maybe a cultural thing..

..where are you guys located?...

USA.

As for the original post. I understand the concept of keeping it from getting to the microphone, but it would appear (at least to me) that if someone can figure out how to keep the wind noise out and still let every other kind of noise in, that somehow, somewhere down the road, an engineer would be able to come up with an algorithm that would do away with just wind noise and allow every other kind of noise to remain in post. I do see the complexity in this; but if we can land on the moon, surely we can get rid of wind noise right?

j razz

Sometimes it can be dazzling how something seemingly so simple is actually vastly complex. Occasionally the analogy of mixing cake batter has been used to describe unwanted sound in this forum. Say you're mixing up your cake and you accidentally put in a sour milk and you don't notice it until you've already mixed everything up. Can you remove the sour milk and replace it with fresh? Absolutely? Separate your cake batter into individual molecules, sift through them all one by one and remove the molecules from the sour milk, keep the rest. Put it all back together and add fresh milk. Simple, right? In theory yes. But the cost and time required to do this would probably be greater than the resources available to our entire planet for generations. So it is possible, but completely impractical.

The same idea applies with wind noise. Once it's recorded it's completely impractical to try to get rid of it. I suppose one could zoom in to the sample level in Sound Forge and pick through with the pencil tool and redraw every single sample the way it would have been if the wind noise hadn't been recorded. Personally, i'd rather change careers to septic system cleaning than attempt that.

The difference with a wind screen is that it prevents the wind from getting to the microphone. The screen doesn't filter the noise; it filters the wind so that there is no noise to begin with. In our cake analogy, this would be the same as sniffing the milk and tossing the sour stuff before dumping it in.

DPA showed their Zeppelin-like "Windshield and Shock Mount" at NAB recently.

I talked to one guy there who said he had used it to record a conversation between two people on a motorcycle doing 60 mph, and there was no wind noise...

Dead cat, Einstein, Fuzzy, Softie are all names to describe a fur-like covering. Blimp, Zepplin, Soda Bottle all describe a hard framed covering. Sometimes you'll see a Zepplin combined with a dead cat.
The DPA is a new type of device that works very well, if you don't mind paying Ferrari-like prices. We have our own similar type of device that also will record conversations on a motorcycle at 60 mph, or more recently tested, a small softie-like covering that can record dialog during a skydive at 120 mph, which is much, much harder to record.
Wind noise can't filtered because it's a dynamic frequency set whether it's hitting the diaphragm(s) of the mic or not. Remember in all the spy movies where the spy turns on water so bugs can't hear clean dialog? That's a factual circumstance. Dynamic noise sources such as wind, moving water, traffic noise, or flying airplanes at close distances cannot be filtered without dramatically impacting the dialog or other subject source. Wind striking the diaphragm(s) of a mic will create a spike as opposed to merely being a constant range of frequencies and harmonics. Bob's explanation is very simple, but very effective.
You can easily make your own windsock/fuzzy/furry/softie/dead cat. Email me and I'll send you instructions. Eventually, I'll do a tutorial on it. I also demonstrate it on the "Now Hear This" DVD, as well as showing various types of devices. Finally, there is also a stream called "Preview 2" that shows a couple of techniques to avoid wind.

All of the salient points about windshields (windscreens) have been made above, so I'll just summarise. Essentially a microphone consists of a diaphram connected to a device that converts movement of the diaphram into voltage. The diaphram is moved by air pressure. Sound waves are variations in air pressure and these variations generate an oscillating signal that we record. Wind exerts pressure on the diaphram. If the wind exerted an absolutely constant pressure on the diaphram then it wouldn't be a problem, because we're only interested in the AC component (of course provided the wind wasn't so strong that the diaphram encountered mechanical limits). However wind is never constant, having travelled over long distances through many obstructions and diversions. So when it hits our mike it is varying in strength and direction, as you know by just holding up a sheet of cardboard. And the dynamic pressure variation in a strong wind is high compared to ordinary sound levels (speech etc). Because the wind has encountered very small (eg leaves in a tree) and very large objects the range of turbulence cells cover dimensions from millimtres to metres, so generate movements in our mike diaphram covering our audidble range (although most of the power is at lower frequencies), so every frequency band we want to keep also contains random variations of wind signal. So if we want to stop wind from affecting our mike there are two ways of achieving this. Put the mike in a box-- stops the wind but problem is that the sound can't get in either, except by vibrating the sides of the box and letting the mike pick that up; of course the wind will also do that. Not workable -- much attenuation. So we need something that will let in the sound we want while simultaneously reducing the wind to a steady constant flow. Screens. Air flowing through a flywire screen is slowed down (each wire creates a wake comparable to the diameter of the wire, converting wind energy into heat). The pressure drop is proportional to wind speed, so fast lumps of air are slowed relative to slower lumps. So the screen tends to even out variations in wind speed and direction. Each screen does a little bit, so if you put enough screens in series the final flow is even and smooth and constant in direction and will not oscillate our mike diaphram. Sound waves, being very small in amplitude and only oscillations of molecules rather than translations, pass through relatively unaffected. So one form of windshield is a mesh sausage surrounding the microphone, and this is covered with a cloth to provide a fine mesh. OK in very light wind. As wind strength and turbulence increases you need more resistance to smooth it out. You can add a greater number of layers of mesh (each separated by a space). To this point in this ridiculously long explanation I haven't mentioned damping, which is also crucial and the reason screens are spaced at a distance from the mike cell. Small scale turbulence rapidly decays (friction and all that) and we want that to happen before the turbulence hits our diaphram. The scale of the turbulence is influenced by wire and mesh dimensions. Hair is very good at damping turbulence, is very porous (lets sound through) and has a very large number of fine cells that offer high resistance to air flow as well as generating very small turbulence cells, and by its nature generates a thick boundary layer of smooth air (so sound generation by the shield itself is reduced compared to a solid box). Foam of course consists of many fine interconnected holes that work in the same way as described above. The stronger the wind the greater thickness of foam needed.
Sorry about that degree of tedium -- I do get carried away!

There are two types of noise you could be referring to. One is a howling from high winds. The other is from normal wind speeds and is often a deep rumble if you're listening on a decent set of monitors. You can't do much with the howling, but you can help the deep rumble type of wind noise.

If it's dialog, I'd suggest using the built in eq and removing (decreasing as much as possible) everything below 250 hz to start. You can go all the way up to 400 hz before it starts to sound awkward.

I've had great success on some projects that were recorded with wind 'rumble' this way.

Sean

------------------------------
broadcast voiceovers

I have to disagree. You can remove rumble with EQ and a high pass, but that's not removal, it's masking or selective cutting. Any wind that is above the rumble freqs, which most wind is, is impossible to remove cleanly regardless of how steady it might appear to be. Wind doesn't need to be "howling" at all, it can be just an odd gust, a light breeze, a serious wind angle, whatever, and it won't be cleanable to any usable degree. It's good you mention cutting rumble, and that's what 99% of wind noise reduction at the camera does; it cuts off low end using an intelligent high pass filter (only sometimes intelligent). If you've got a gust that hits with similar characteristics, you can also use a multiband compressor targeting just those freqs, but noise reduction won't get it out well.
Jeffrey Fisher covers all of these examples really well in his Noise Reduction DVD, FWIW.

Yes turbulent wind does make sound of its own, which is caused by vortex shedding (eg. powerline sound). That is part of the background noise. However noise generated in the microphone, which you refer to as rumble, can be prevented by windshields. If the shield isn't adequate then some interaction still occurs and this tends to be at the lower end of the specrum. Yes I've used SR in Soundforge to remove this "rumble" but 400Hz is well into normal speech range. I didn't care for the result even though the speech was intelligable.

However they do it, it's amazing when you see reporters covering a hurricane. In most cases, you don't hear any wind.

Sometimes when you have recorded in stereo the wind may have attacked the two microphones differently. It may be worth your while to check if the one channel alone sounds better than the other. (In Vegas: rightclick the event and select Channels > Left or Right only.) I have been saved by this a few times.
Tor

Geez, I am always adding wind noise to make it sound real. You can use the clipped peak restoration to reduce severe rumble.

JJK

Today I came across some audio that had terrible wind noise...and you can give a listen to what some bass reduction can do. It doesn't entirely remove the wind, but it sure makes it sound much more usable.

The audio was done with a Sennheiser MKH-416 with the wind sock on it. Original plays first, then the fixed one. In vegas, I dropped out everything below somewhere near 150 hz.

mp3 posted here



Sean
------------------------------
broadcast voiceovers

Certainly for that kind of problem some bottom end roll off can work wonders. But where we were at before was trying to remove wind noise from mics without a wind sox.
In your case the wind sox has prevented the howling sounds from the air passing over the cavities in the mics body, get those howls and whistles mixed into your audio and it's a different matter entirely.

As a matter of interest what kind of wind sox did you have on the mic?
My guess is that the full Rycote Blimp and Dead Cat give the best results. They cost heaps but included good isolation for the mic and that could also be important. In a good gale the wind hitting a boom pole will create vibration in the pole, without isolating the mic from the pole that'll end up in the recording. Also I think the blimp helps by getting the fur off the mic body giving the air space to settle.

Bob.

The K-Zep looks like a good system as an alternative to Rycote and cheaper (although I haven't checked that the kits are equivalent in items).
edit: they changed the page reference, but the item will cycle into view.

windshield system

I know this thread is long gone but I just came across this mic while perusing the latest Canford catalogue...

Audio Technica Adaptive Array Mic

The blurb states it is "a mono directional microphone that combines analogue and digital signal-processing to reduce unwanted sounds, including wind-noise, while maintaining a natural sound of the target sound source".

So perhaps you can filter it out after all. It's a snip at £1650 (plus VAT) if you want to try it out!

Cheers

gb

Well there's also the new 'Blimp' from DPA which not only does better at reducing wind noise than a dead cat but barely touches the high frequencies. It is expensive but not overly so compared to the traditional fare from Rycote and the DPA Blimp is shower proof, one can even get a raincoat for it.
So I guess the next challenge having killed wind noise will be killing rain noise!

Bob.