The Yandex Cloud which is a cloud platform, introduced the Yandex SpeechSense speech analytics service that uses the capabilities of YandexGPT. The technology allows a detailed analysis of calls and chats of contact centre operators, including assessing their tone and involvement in dialogues.
With the help of Yandex SpeechSense, companies will be able to improve the quality of employee experience in voice channels and reduce the cost of communication analytics. The service is available to platform users in Private Preview mode and is available upon request.
In SpeechSense, speech analytics occurs in several stages. First, call recordings, chats, and data from CRM systems are uploaded to the cloud. There, the speech is converted into text using Yandex SpeechKit recognition technology. The resulting text content is automatically checked for compliance with the script. For example, it is noted whether the operator greeted the client or whether there were negative statements in his speech. After this, the transcript is sent for a more comprehensive analysis to YandexGPT. The neural network can work with complex metrics and determine, for example, the tone of the dialogue, the level of empathy and involvement of the operator, and assess customer satisfaction.
Grigory Atrepyev, Product Director at Yandex Cloud:
Yandex SpeechSense will help businesses make the most of insights from voice communication channels. Several Yandex services and our clients have already tested SpeechSense on real tasks. Now we are opening access to more users.
Integration with YandexGPT provides more opportunities for speech analytics in the service. Currently, SpeechSense uses a general neural network model. The developers of the cloud platform plan to further train and improve it so that it can, for example, check operator responses for compliance with the editorial policy of the service and business processes.