Noise cancellation with Audacity
by Steve Smith

Following on from Ben’s excellent forum post about listening to .psdata files, I wanted to share the noise-cancelling feature in another audio software suite called Audacity, which you can download here.

Ben has referred to Sonic Visualizer in his forum post, which is an excellent audio software suite we both use for heat maps/spectrum views when interpreting noise.

The reason I mention Audacity is because of this video by Dusty Porter, which describes how to remove unwanted noise from an audio capture.

I am sure Sonic Visualizer has the same facility, but for now, I want to share how the noise cancellation feature in Audacity could be helpful with NVH.

To make a long story short: I was sitting on an airport bus listening to its transmission/differential whine as we were transferred to the terminal.

I took out my mobile phone and recorded approximately 40 seconds of the whining sound (I know, I should get out more).

I had to convert the format of the audio file on my mobile to .wav format so that I could import it into the NVH software. I have described the process in this forum post: Audio Import from Smart Phones.

Here is the mobile phone recording converted to .wav format: Bus Whine uS.wav

If you use headphones to listen to the file, you can hear the whining sound between 0-10 seconds and between 27-37 seconds.

This is what the whining sound looks like when imported into the NVH software:

In the image above, we can see that when the whining sound is present, we have a clear peak of 375 Hz. This could relate to a tooth contact frequency within the transmission/differential. To confirm this, we’d need the transmission ratios, tooth counts, tyre size and road speed, all of which I didn’t have as this was not a customer complaint and I was meant to be in holiday mode when I recorded it!

What do I mean by tooth contact frequency?

Think of a differential crown wheel with a poor or worn “finish” on each tooth. Each time a crown wheel tooth makes contact with a pinion tooth, noise may well be generated. The speed/frequency of the crown wheel will be T1 (wheel speed). If we assume that our crown wheel has 47 teeth rotating at a frequency of 8 Hz we get this: 8 Hz x 47 = 376 Hz tooth contact frequency.

While tooth contact frequency requires both rotational speed and tooth count (as demonstrated above) my calculation is, in this case, purely hypothetical.

So how can Audacity help?

Let’s create a reference waveform of the above capture by using the right-click feature in the Frequency view.

By using Audacity, we can remove the part of the audio where we can’t hear any transmission whine, between 11 and 26 seconds. This part of the audio is not required and hinders the identification of our offending frequency.

These are the steps to take:

  1. Download the file “Bus Whine uS.wav” above.
  2. Open Audacity, click on File and open the “Bus Whine uS.wav” file.
  3. Left-click on the imported waveforms in Audacity and drag from 11 to 26 seconds.

You can now use the selected area above as “unwanted noise” and remove it from the entire capture via the Noise Reduction feature.

Click on Effect > Noise Reduction… > Get Noise Profile.

Audacity will now learn what to filter as noise from the full audio capture.
N.B. When you click on “Get Noise Profile” above, all the noise reduction windows will close.

Now select the full waveform (use the Ctrl+A). Click on Effect again, but this time you select Repeat Noise Reduction.

The noise captured in the selected area of Image 3 will now have been removed from the entire capture. Select File > Export > Export as WAV and save to a location on your PC.

The following screenshot is the mobile phone recording edited in Audacity and imported into the NVH software again. Remember, this is the same Audio as in image 1 but with the noise-cancelling applied: BUS WITH NOISE CANCEL.wav

Now we can compare this to the reference waveform created in Image 2.

As you can see, the offending peak (present when the whine occurs during playback) can be seen with a defined ramp either side of a clear peak at 375 Hz. The playback through NVH is the real winner here as the whine is far more pronounced and easier to define when using headphones.

The above is only one example of how Audacity is a useful program to use in tandem with the NVH software. Other examples could be cabin noises, where you can mark and remove noise from cabin boom, tyres, etc. to enhance and reveal the customer’s offending noise (squeak, creak or whine).


Add comment

Your email address will not be published. Required fields are marked *