Noise can be very difficult to cancel especially in crowded environments like call centers and shared workspace. The noise in such locations has two main components:

1- Office sound and static ambient noise

2- Clear voices of other people talking around you

While the first component does not sound like human voice, it can be canceled via an AI based noise cancellation algorithm that is trained to differentiate between human voice and non-human sounds. On the other hand the second component requires a methodology that is based on something like a Noise Map to identify the location of the human voice and decide if this is the primary speaker or not.

As can be understood in order to effectively cancel the noise in these environments you should use two noise cancelling methodologies in a sort of a pincer movement.

A pincer movement, also known as a double envelopment, is a military maneuver in which two forces converge on an enemy from different directions, effectively surrounding and trapping them. This tactic has been historically used in warfare and can offer significant advantages in the battle field. However, executing a pincer movement effectively requires careful planning, coordination, and timing. If not well-executed, it can expose the attacking forces to counterattacks or risk being counter-enveloped by the enemy.

In our case a similar approach can be taken by carefully and accurately combining two noise cancelling methodologies on the noisy audio signal. The result will be a much cleaner audio, improved voice quality and customer experience.