The supercomputer trains to recognize emotions and improves the work of call centers

“Call-centres use tone of voice emotion classification to improve quality of the service. This feature is unique selling point for pitchptatterns.com software”  Evalds Urtans, CEO, asya.ai.

Asya.ai has developed pitchpatterns.com, which is probably the best and most effective software that call centers can use to analyze the quality of incoming calls and automate various processes.

Challenges & Solutions

Classification of emotions by tone of voice is difficult due to the lack of public datasets and lack of open-source models. It also requires significant hardware resources to perform hyperparameter searches and train your model.

In cooperation with the RTU HPC Center, we trained our emotion classification model to detect Happiness, Anger, Sadness, and Neutral emotions in the tone of voice. It achieved 95% accuracy.

Asya.ai managed to create a unique service for companies that work in the field of customer service daily, and calls from various customers, including dissatisfied customers, are an integral part of everyday life. By quickly recognizing the customer’s mood by the tone of his voice, specialists working in call centers can continue communicating with the customer according to the algorithm developed by the company itself, which is how to communicate with the customer in certain emotional states.

With pitchpatterns.com, a product developed on a supercomputer, customer service companies can use artificial intelligence to manage call center operations, ensure quality control, and change the usual (traditional) approach to customer service over the phone.

Benefits

– Deep Learning model training

– New feature and unique product feature that improves sales

An example

  1. Emotion classification by tone of voice before the HPC project: 52% accuracy
  2. Emotion classification by tone of voice by the HPC project: 95% accuracy

Such large industry companies already use the quality control service pitchpatterns.com as SIA “TET”, SIA “inBank”, SIA “Bono”, SIA “Rīgas ūdes”, CSDD, Latvian State Radio and Television Center, SIA “Aver”.

For more information: Evalds Urtans, SIA ASYA, [email protected].