Image
Image
Quality monitoring with speech technology
Quality monitoring with speech technology
At Telecats, we develop speech technology for contact centres. More and more often we get the question: “Can you also say something about the caller's "emotion"? Are our customers satisfied after the conversation or not? And can you indicate whether the conversation between customer and contact centre employee is going according to plan or whether someone should intervene?

With language and speech technology, we can increasingly detect emotions in speech. The basic emotions of fear, anger and happiness are similar in all cultures and therefore easily detectable with software analysis. It is more difficult to detect things like irony and sarcasm because they are more subtle and differ from one culture to another. But silences in a conversation, crosstalk and raising the voice are emotion markers that we can capture very well.

Video: example of Quality Monitoring


In addition to analysing the sound signal, we can also recognize what is said with speech recognition. In this way, we know what is being said and how it is being said. We can also see how long people pause for the next word before they speak. We look at "what" someone says and "how" they say it. If things are in danger of getting out of hand, we see that people no longer allow each other to speak and therefore interrupt one another. They are also going to talk louder and use certain words. With the combination of all these metrics, we can indicate fairly accurately that the conversation is in danger of being derailed. The same approach is used to determine whether callers are "satisfied" or not. This is not very simple, but it turns out that the average the computer calculates and what some people calculate, is close together.

Video: example of Quality Monitoring

In addition to analysing the sound signal, we can also recognize what is said with speech recognition. In this way, we know what is being said and how it is being said. We can also see how long people pause for the next word before they speak. We look at "what" someone says and "how" they say it. If things are in danger of getting out of hand, we see that people no longer allow each other to speak and therefore interrupt one another. They are also going to talk louder and use certain words. With the combination of all these metrics, we can indicate fairly accurately that the conversation is in danger of being derailed. The same approach is used to determine whether callers are "satisfied" or not. This is not very simple, but it turns out that the average the computer calculates and what some people calculate, is close together.



Real-time Dashboard

On the basis of this data, a real-time dashboard can be created where all on-going customer contacts can be displayed in addition to the basic conversation data, as well as the course of the emotion in the conversation. This can be used for training and support by supervisors. A supervisor can start monitoring and support an employee on the basis of declining sentiment during a conversation with a customer.
Image

Share this Post