Background

Call centers suffers from significant ambient noise that is originated from multiple locations like:

• Other agents talking in the background.
• The supervisor/team-leaders shouting to instruct or enthuse the agents.
• A nearby gathering in the coffee corner.

Such noise reduces the quality of the calls and directly impacts customer satisfaction. In addition the existence of the noise reduces the accuracy of call center speech analytics. In this post we will discuss how you can fight this noise and also keep a close eye on its behavior.

Know your enemy

How good do you know your enemy (the noise in the call center)? Taking into account the acoustics in the call center, the physical location of each agent and the equipment that is being used (e.g. headsets), can you tell what is the origin of the most disturbing noises for each call? What happens when the layout in the call center floor changes? When few of the agents take a break, how does this influence the identity of the disturbing noises for all other agents?

It would probably help if you could visualize this using some virtual map.

The virtual map we need is not the “world’s map” but a Noise Map of your call center. The main requirements from such a map are:

• For each phone identify the most disturbing sources of noises and provide some scoring on the level of noise.
• Take into account temporary changes in the call center, for example when few of the agents are on a break.
• Continue learning and updating the map to take into account any changes that happen in the call center.

Surely this kind of information can be of help when you try to attenuate the noise.

So, the question is, how can this map be generated and maintained?

The technology

Noise Firewall is a plug-and-play software solution that is equipped with advanced reference-based noise reduction (RNR) technology. Once installed, the Noise Firewall automatically connects to the multiple phones in the call center and starts processing a sophisticated correlation between the multiple audio streams. Such correlation enables the Noise Firewall to identify the primary agent in each call and actively attenuate the ambient noise. As a result, the Noise Firewall will actively, remove background noise in the call center. In addition, the Noise Firewall will build a real-time Noise Map of the call center. You can use this map to further improve the behavior of your call center.

The Noise Map

The real-time Noise Map is built and maintained by the Noise Firewall based as the analysis provided by the RNR algorithm.

In this map, each row contains a subject agent and a sorter list of the most disturbing agents for this subject agent. The darker the color, the higher the noise. For example, the first row in the above map, shows that currently for calls made by agent JDow, the most disturbing noise is coming from agent RSmith and additional source of noise is coming from agent QMurray. This map is constantly updated based on the aggregated data that is detected by the Noise Firewall.  In many cases, the above Noise Map will have resemblance to the call center floor map as illustrated below.