k y m a • t w e a k y — the kyma collective || view the current website || February 2015 Archive

/ WebHome / How.MicrophoneWindNoiseFix


How Section

How do I...
Add a question...


All Sections


Login / Register 
Change password
Forgot password?

Symbolic Sound

Kyma Forum
Eighth Nerve

TWiki Links

TWiki Formatting FAQ
What is TWiki?
%SESSION_ENDIF% twiki.org

Question (or Task)

I know this is not, strictly speaking, a Kyma-tech issue, but someone here may have a Kyma-related solution. I'm looking for suggestions on cleaning up mic recorded outdoor audio with lots of wind noise. Worst offending element is probably the low frequency component. I was thinking of going with something related to noise cancelling technology where I would gate an inverted signal of the audio with the volume envelope of the original. But initial results have been "disappointing" to put it mildly. Perhaps I'm doing something wrong ... or the extra processing results in the two audio files no longer being perfectly out-of-phase by the end of the processing chain? I suppose compression may also work, but so far it's proved ineffective. Curiously, the best results so far (though not still not very good) have been with Sound Soap (the cheap version) monitoring the noise sideband rather than the signal!

-- MarkPhillips - 28 Feb 2005


Hi Mark

Noise cancelling is probalby a non starter. It works in headphones where you have two signals , one with the noise only and another with exactly the same noise but with a signal as well. You cannot cancel when you only have one signal in the first place. You could adjust the signals level with compresion and subtract this from the original, but that is just the same as if you had done expansion on the original signal in the first place. Expansion or gating is the key but as you have probably found out when the gate opens the noise comes through as well. So what you realy want to do is to seperate all the frequencies and gate each one seperatly and then join them back together. This could be done in kyma but you always get problems where the signals components that sits on the join between bands gets transformed into a phasey type of sound when some bands are removed and the bands are not joined up in the same way as they were pulled apart. This is the techneque that soap uses and it is a compromise between this phasey noise and the amount of cleaning that can be done. Also all the gates need to have there thresholds set and this can be automated by sampling a noisy bit (with no signal) and using the senced levels as the thresholds. But wind noise is even harder to remove than ordinary noise as it is not a constant sound. I suspect that the makers of soap have done a lot of experimenting to get just the right settings and that to do a similar job in kyma would take a lot of time before you could get any where near the results that you can get with soap. If you do want to try maybe David McClains? Zilloin band compressor could be transformed into a Zillion band gate but the results are not gaurenteed.

I wish I could be more optomistic.

-- PeteJohnston - 28 Feb 2005

Pete -- Thanks for the response. It's more or less what I thought, but I figured it was worth at shot. Outdoor/location recording is not a part of my normal realm. And in fact I didn't record this audio, either -- it was sent to me. But since this situation must be encountered frequently by a significant segment of recording engineers and sound designers, I hoped that maybe there were some tricks out there that I didn't know about.

Of course. I misspoke when I mentioned "noise" in connection with "cancellation" ... since I don't have a separate noise signal. What I was referring to was an attempt to use a controllable amount of total signal cancellation (using an inverted original signal) that would be gated to kick in when the wind noise was most obnoxious. Of course that would process would knock down the ampltude of everything, but that would still be be a lot better than what I have (since at those moments all you can hear is the wind buffeting the mic diaphram.) I wondered if this idea might produce fewer digital "twinkie" artifacts than some other kinds of noise reduction schemes and algorithms I've used in the past.

Regarding a steady noise versus a dynamic one like wind, I did find it surious that Soap thought the wind noise was signal ... and most of what I wanted to hear instead was noise :-)

-- MarkPhillips - 02 Mar 2005

Question: How do I "cancel" mic recorded wind noise?
Keywords: Noise Cancellation Amplitude Follower Compression Expansion

© 2003-2014 by the contributing authors. / You are TWikiGuest