Sound Cancellation Testing

Rick Guyton · July 18, 2019

During KazooCon, Yealink and Polycom both promoted their sound cancellation features. I decided to do some testing on these features to see just how effective they are. I thought I'd share my results and testing methodologies with you all.

Testing Methodology

I created an audio file that included music, conversations, kids playing and dogs barking. I've included it here for you as TestBackground.wav.

Then, I made custom configuration profiles for each setting I wanted to test. Yealink only has 4 (Off, 1, 2 and 3). While Polycom says there's no limit to the size of the variable you can use in their threshold. I noticed that 8 was the default level. And after searching around I found some people setting it as high as 150. So, I used 8, 10, 20, 50, 100 and 200 for my tests. For each of the profiles, I also changed the line name so that I could insure that my custom config was applying correctly. On the polycoms, I had to include my entire account reg line to do this, so I've stripped that data out. But, otherwise, you can download the custom files I used below as well.

I played the sound at my computer and put the phone about 12' away and 4' down from my speakers. I don't have a decibel meter or anything, but the speakers were almost all the way up. And I'd say that the music was very loud, the conversations sounded like being in an extremely busy convention center, the kids sounded like they were in my office playing and the dog was about as loud as if it was barking in my face. If that helps... Very loud that is to say.

Then I made a callflow that dumped to just playing silence and setup call recording to capture the results.

After doing this I noticed that every single test category failed to completely mask the background sounds at the very high level I initially set them at. So, I also did another round for both phones on their highest cancellation level and a lower background noise level.

My test devices were a Polycom VVX410 and a Yealink T53

Results

I have to say, even as a Yealink guy, I was very impressed with the Polycoms. And, less so with the Yealinks. I struggled to notice really any difference with the Yealink acoustic shield on vs off and certainly could find almost no difference between the different levels. There doesn't appear to be any down side to enabling the cancellation on either of the phone, even at the highest levels the voice didn't sound distorted. However, I did notice that at higher levels, the Polycoms sound like they are muted when the operator isn't talking. Is that a good or a bad thing? I'll let you decide for yourself, but I like it personally. I've included my test result recordings here as well so you can listen to them it you'd like.

When I tested at lower background volume levels, I did notice that the T53 did perform better than I expected even with the acoustic shield off. This might be due to the new microphones that they were talking about at the convention as well. But again, I struggled to notice any difference between acoustic shield being off and on.

At the end of the day, as a Yealink guy, this VVX 410 is staying on my desk and setup with Acoustic Fence & threshold set to 200.

A Few Notes

I'd say the best test hands down was the VVX 410 with softer background noise and a threshold of 200. But, even then, if you listen carefully, you can still hear my kids a little bit. 200 isn't the maximum threshold. I wonder if it would get even better at a higher level? I'd be interested to know if anyone tries it.

Acoustic Fence seems to be supported since F/W 5.3 if enabled by provisioning. However, the setting is not available in the UI in any version that's supported on the provisioner. I upgraded to the newest F/W through the web interface on the phone. After that, I could access the setting, but only in the phone's menus, not via the WebGUI.

Hope you find this valuable! If someone wants to duplicate my testing, I'd love to see if you find similar results. I've included everything you need to fully duplicate the test here.