documentation for windnoise algorithm ?

Hello James,

The wind noise reduction algorithm that was recently added to the SigmaStudio library is different than what was mentioned in the 2002 article you referenced. This algorithm is used to detect the presence of wind in a signal (typically the input is from a microphone) and smoothly enable/disable a high-pass filter. Since the sound energy of wind is primarily at low frequencies, a simple filter is sufficient to remove much of the wind noise from the recorded signal.

The wind noise block in SigmaStudio takes a stereo input from two microphones and outputs the processed stereo signal on the top two of the three output pins. The bottom output pin can be left disconnected. There are three parameters on the block that will be of interest for tuning the algorithm.

• The frequency control sets the detector filters. This parameter should be set as low as possible so that wind noise is removed, but the desired audio signal is preserved. The frequency parameter should be tuned while the system is presented with a constant wind noise, such as from a fan blowing across, not directly onto, the microphones. The value can be entered by clicking the up/down arrows or by entering text directly in the box

• Level 1 should be tuned while turning the wind source on and off and simultaneously tuning the parameter setting between 0 and 100. The Level 1 setting is recommended to be between 60 and 90, but this will vary depending on the application. The value can be entered by clicking the up/down arrows or by entering text directly in the box.

• Parameter Level 2 should be tuned in the same way; its settings range from 0 to 15, with 0 being for strong wind noise and 15 being for a signal with a weak wind noise component. The value can be entered by clicking the up/down arrows or by entering text directly in the box.

We apologize that the description of this algorithm is not currently included in SigmaStudio's help file. Sometimes the documentation for the new processing blocks lags behind the actual algorithm implementation by a few weeks. This block will be fully-documented in an upcoming version of SigmaStudio.

Hello

thanks for the reply.

No apology is necessary, I understand that it takes time to document advances in software.

no specific question below, nor expected reply, just related ramblings, comments and guidance is appreciated though

I'm guessing that due to the high pass filter/mic input that this algorithm is most beneficial when using a mic on a person speaking or possible singing.

What I was hoping for is something that would use a mic to register wind/road noise in a vehicle, then apply an algorithm to cancel the impact of that noise on the stereo output.  I guess maybe inject the noise at 180 deg out of phase into the stereo input.  I wonder and dream about the ability or successful execution of something like that.

My car, a convertible , came with a factory external amp.  Well really it used the head units powered output as its input so I would more readily call it a power booster/sound processor.  Anyway it had a couple inputs from the vehicle.  If I recall one was the speed of the vehicle, and the other was whether the convertible top was up or down.   The response of the booster/processor apparently varied based upon those inputs.   I'd like to implement a variation of this, possibly using the vehicle supplied outputs.  Well I was gonna make a reference to adi mems Accelerometers , which I had seen in the digikey catalog, but the link is down.

rperkins

You're correct that this wind noise reduction algorithm is intended for voice applications, such as mobile phones, handsfree car kits, Bluetooth earpieces, etc. Wind noise can easily overwhelm the voice signal in these applications, making the person speaking very difficult to understand. Removing the wind noise makes this speaker's voice more intelligible to the listener on the far side of the link.