How can echo suppression badly impact call quality ?
- On January 16, 2016
- echo suppression
Background
Can phones/soft-phones with built-in echo suppression badly impact the audio quality? Surprising as it may sound the answer is positive. In other words, when a call center, or any other organization, purchase phones with built-in echo suppression it is jeopardizing the overall call quality. In order to analyze the reason for this phenomena we first need to understand few basic concepts in audio processing.
Echo Suppression
Echo suppression is an old and simple method of handling echo during phone calls. The logic of this method is very simple: during a phone call there is usually one person talking while the other person is listening. So, if we force the call to be half-duplex (i.e. only one side is allowed to talk at any given point in time) the echo can usually be suppressed. Half-duplex mode is like a walkie-talkie: while one side is talking, the audio that is generated on the other side (that includes the echo) is suppressed and not transmitted.
Because this technology is very simple it is usually implemented in low-end phones and softphones. If this technology suppress echo, how can it harm the audio quality?
Continuity of the Audio
The echo suppression breaks the most fundamental concept in audio processing – the continuity of the audio. As an example, let’s assume we have a phone in a call center that is using the echo suppression method. While the agent in the call center is talking, his/her voice including the background noise is transmitted to the customer on the other side. But, when the customer starts talking, the audio that is captured on the agent’s side is suppressed and usually silent is transmitted instead. So, if you are to analyze the audio that is coming from the agent, sometimes you will hear the agent’s voice including the background noise and sometimes you will hear silent. Apart from the fact that this behavior irritates the human ear, what will happen if we want to remove the background noise? How can a noise reduction algorithm accurately learn the nature of the ambient noise that disappears every few seconds?
The solution
Try using equipment that its built-in audio processing does not break the continuity of the audio. For example, instead of an echo suppression algorithm, use an echo cancellation algorithm that maintains the continuity of the audio. If using such phones or softphones is not possible or too expensive, we recommend using simple phones (even basic analogue ones) that do not do any change to the audio. Professional audio filtering for noise reduction and echo cancellation can then be done in the network, for example using PBXMate.