So for the past few months have been have been investigating microphone wind noise. We choose microphone wind noise as this came very high in our online survey into the main issues that can degrade the audio quality. The survey is ongoing so do please take the time to carry it out if you are interested.
To investigate microphone wind noise, the first task is to understand how it is generated. Luckily has already been significant research that has already been carried out to this aim, so a thorough literature review was carried out. The dominant source of wind noise in outdoor microphones are turbulent velocity fluctuations in the wind which interact with the microphone and are converted to pressure fluctuations. There are other less significant factors which can contribute, for example when the microphone is embedded in a device and this is placed in a flow, this can cause vortex shedding and other resonant type behaviors.
To understand how wind noise is perceived a large database of examples is required. The idea of recording large amounts of microphone wind noise outside is problematic as there are always other sounds present. This could bias any tests. It was therefore decided that a microphone wind noise simulator should be built. Using a microphone wind noise simulator it will be possible to generate audio for a range of wind conditions which has no other sounds present. Other audio could then be mixed in afterwards simulating the presence of other sounds. Two wind simulators have been designed the first is based on real microphone recordings with an unshielded microphone. The second is based on a model developed based on what was learned from the literature review about how a shielded microphone performs.
Recordings of microphone wind noise were made in the University of Salford acoustic test facilities’ silencer test rig. The silencer test rig is used as it has been designed to minimize of fan noise, this means that the main source of noise at the microphone is wind noise.
A sonic anemometer was used to measure the measure the air speed at the end of the test rig. (A sonic anemometer is a device which measures the wind velocity at a point using ultrasonic pulses). A minutes worth of noise was recorded for a range of different ‘wind speeds’. As one of the things which characterizes wind noise is it’s gustyness, or non-stationary, real sonic anemometer data is used and a dynamic mixer is implemented which selects short sections of real microphone wind noise audio depending on the wind velocity at that moment. This creates a time varying noise which is representative of microphone wind nosie but without any other sounds present.
The audio for the second model is generated by randomizing the phase for a known power spectrum, this generates a stochastic instance of noise with the desired spectral character. The same dynamic mixer is used to create the time varying simulation.
Here are two audio examples, one generated using a simulation without a wind shield and one with, both used the same wind velocity time history to generate the audio.
Comparing the two simulations, the shielded is significantly quieter than the unshielded simulation, which is expected, but additionally there is some audible resonances on the un-shielded simulation which is caused by vortex eddy shedding as the wind flows past the microphone.
Now that a flexible model of microphone wind noise has been defined, there are two research paths that we will be exploring over the next couple of months. Firstly, how different parameters such as wind noise level effects the perception of quality, and secondly can we make a wind noise detector which is robust to the presence of other signals.