Correcting "Wind Phasing"

I’m interested in removing “Wind Phasing” from outdoor concert recordings. As I understand it, gusty wind degrades the high frequency content, giving a dull sound. But, when the wind pauses for a second, the high frequencies return and you get clear-sounding audio. The fluctuating levels of high frequency content give a so-called “swirling” effect.

I’m thinking that it wouldn’t be that hard for a skilled someone to create a plugin that would dynamically control the high frequency content. Using a “lookahead” algorithm, the plugin would dynamically boost the high frequencies, where needed, to compensate for them being blown away. Might need some rules to ratio the HF boost to the mid and bass frequency strengths (so you’re not over-boosting high frequencies when the song gets the the quiet part…), etc.

Does this already exist? Anyone care to take a stab at it? I’d be happy to offer sample files and encouragement!

wind degrades the high frequency content

I haven’t heard of that, but I have heard of someone trying to record music without turning voice processing off. Voice processing (echo cancellation, etc) hates music. Voice Processing gives music a swirling, honky effect, degrading the high frequency content.

Very few people walk away from recording a music concert with a workable show. Everything is stacked against you. The Big Kids use special microphones and accessories—or get a feed from the house mixer.

I’d be happy to offer sample files and encouragement!

Possible, but as a rule, the programmers/developers have all they can do to keep Audacity running.

Koz

The effect you describe is basically a “multi-band compressor”.
The phenomenon that you describe is due to “auto gain control” (AGC) in the recording device fighting with sudden blasts of low frequency noise. The “fix” is to dramatically cut those sub-sonic blasts when they occur and boost everything else. It is very difficult to do this effectively, partly because the AGC is pushing the audio that you wish to recover down into the noise floor, and partly because the speed and amount of auto gain adjustment during recording is unknown (Ray Dolby built a career from being able to accurately match frequency dependent compression and expansion, but he had the benefit of designing both ends of the system, not just the “playback compensation”).

The professionals get round the problem by either, using a live feed (as koz described, or wrapping the mic(s) in “blimp” and/or “dead cat” (http://www.rode.com/accessories/blimp), or by recording the audio separately in a studio and mixing the recording and video later.

There are some free VST multi-band compressors available, but I don’t use VSTs so you may need to search the Internet - don’t expect great results from this type of post-processing, professionals avoid the need with very good reason :wink:

Thanks for the replies, but both miss the mark.

Steve, I think you are describing standard low frequency wind noise or “wind rumble”. Yes, that’s common unless windscreens, typically fur on top of foam, surround the microphones. I’ve got that covered and have no low frequency wind noise. Auto Gain Control was not used on these recordings, as tapers know (or learn) to avoid using AGC.

I’m just interested in getting my hands on a plugin or Nyquist script that will dynamically adjust the high frequency content and keep it in line with the mids and lows (which are steady and fine). If I had a high frequency knob to turn, I could just about do this in real time since the fluctuations in high frequencies are easy to hear. However, I would always be “late” with manual adjustments since I can’t anticipate the changes the way a “look-ahead” algorithm could.

if you’re not talking about the effects of “wind blast”, then how does the weather affect the recording? Are you recording at a great distance from the stage? Perhaps you could post a link to a short sample to illustrate what you mean.

I think some of the tools used in the ambisonic sphere could fix some of this effect. I’ve read papers about compensating for it this way when playing, not after the recording has already been made. And there won’t be any ready-made tools, afaik, because all of this is very experimental.

But, you can’t recreate what isn’t there. If the high-loss is severe, there’s no way math can rebuild the lost content.

You could say that the earliest experiments trying to solve this go back to 1973, when the Grateful Dead built the “Wall of sound” to remedy this effect, with some success. The system was so big that it wasn’t very practical and it was only used for two years. See:

https://en.wikipedia.org/wiki/Wall_of_Sound_(Grateful_Dead)

Here’s an audio sample of what I’m calling “wind phasing”. Notice that the low-end frequencies of the guitar stay relatively constant. However, the high-end of the guitar, and especially the drums/cymbals, fluctuate quite a bit.

https://dl.dropboxusercontent.com/u/19036522/wind%20noise%20example%2016%2044.1.wav

I tried a very simple 3-band compressor on the high-end only and the result was worse. Maybe a more sophisticated compressor would do the trick? Maybe a more sophisticated audio processing program is needed?

Thanks for your help!

I’m guessing that this is a “bootleg” recording, so getting a direct feed from the mixing desk, or getting closer to the stage are out of the question. If that’s not the case, then the solution is to either get a direct feed from the desk, or get closer to the stage, or both (and synchronise them later).

Or using long shotgun mics :mrgreen:

Shotgun mics don’t help because the problem is caused by turbulence in the air.

The speed of sound varies depending on temperature and pressure, so when there is a large mass of turbulent air (“wind”) between you and the sound source, the sound waves become distorted and do not propagate at a constant speed or direction. High frequency sounds have a short wavelength (about 10 cm at 3.5 kHz), so it only requires tiny changes in speed for high frequencies to “catch up” with sound that has been “bent” differently (due to the turbulence) and cancel itself out. (the time period for half a cycle at 5kHz is about 0.0001 seconds).

If you ever go to outdoor concerts, the effect is often very noticeable when listening from outside of the main concert area.

Yes, Steve, I know. Hence the smiley…

I also don’t believe any plugin can fix it. But who knows? Maybe some sort of synth? :confused: