All news

Yandex Cloud introduces neural network for business capable of analyzing emotions by voice

The combination of several machine learning models will allow for the recognition of more nuanced speaker emotions such as uncertainty or sarcasm

MOSCOW, March 11. /TASS/. Yandex Cloud presented a neural network capable of recognizing emotions, negative, informal statements, and obscene language via speech, as well as the speaker's gender. According to the company, businesses can use it to develop voice assistants and virtual call center operators.

"The empath neural network was built using Yandex SpeechKit speech recognition technology. With its assistance, businesses will be able to construct voice assistants and virtual call center operators who understand human emotions. This will enable businesses to increase the quality of phone conversation analytics as well as better adjust the work of call centers. centers for each client and swiftly respond to emergency situations during the conversation," the company said.

In the near future, the new model is expected to be used in SpeechSense, a voice analytics service that uses Yandex SpeechKit technology and is integrated with the YandexGPT generative neural network. The combination of several machine learning models will allow for the recognition of more nuanced speaker emotions such as uncertainty or sarcasm. Furthermore, neural networks will be able to determine whether the operator was truly engrossed in the client's situation, whether they were attempting to help solve the problem or simply ending the conversation as quickly as possible, the company noted.