Yandex SpeechSense

The main articles are:

• Speech Recognition (Technology, Market)
• Speech technology: On the path from recognition to understanding
• Data mining Data mining

Yandex SpeechSense is a service for deep analysis of communication channels (voice and text).

2026: AI assistant to check quality of service in contact centers

Yandex B2B Tech on February 5, 2026 presented the ability to create AI assistants to analyze dialogs between customers and contact center operators. They will help you quickly assess the quality of service and identify possible problems in communication. In large companies, digital assistants will save hundreds of hours when processing several thousand dialogs a day. In two months of testing, more than 30 companies have already launched about 700 assistants. You can create AI assistants for your tasks in the Yandex SpeechSense voice analytics service.

AI assistants can automatically assess the level of customer satisfaction, compile a brief summary of dialogues and give recommendations on how to improve the quality of service. They can also be used to analyze offline dialogs at points of sale, to evaluate HR communications and other tasks. Companies can create several assistants for different purposes at once, and they can also use ready-made template assistants: for example, to analyze active sales or control customer technical support.

We strive to make communication analysis quick and accessible to all our customers. With the advent of an AI assistant in Yandex SpeechSense, any user can get insights from dialogs by simply specifying their request to them. In this way, we expand the scenarios for using communications analytics. For example, the product manager can quickly understand which elements of the application most often raise questions from users, and the sales director can check which stocks and offers work and which do not, "explained Elena Belobrova, head of business development at ML services at Yandex B2B Tech.

2025: Voice Analytics Service Optimization

The Yandex Cloud cloud platform has updated the Yandex SpeechSense voice analytics service. Now the business can analyze records of dialogues of employees and customers in bank offices, stores, clinics and other organizations using neural networks. Thus, the business will be able to increase sales efficiency and service quality.

The Yandex SpeechSense update expands the capabilities of marketing analytics: the business will be able to determine customer loyalty, the effectiveness of marketing offers and find the most successful examples of working out objections. This will also allow a more comprehensive assessment of the employee's work, for example, to determine whether customers like his communication style, which words and phrases should not be used, and which, on the contrary, can lead to a deal.

According to Yandex Cloud estimates, the potential volume of the offline voice analytics market is 11.5 billion rubles, which is more than 320 thousand recording devices, including audio pages and voice recorders. This figure is approximately 1.9% of the global market volume, which reaches about $6-7 billion. Among the key customers of voice offline analytics are banks and financial companies (they are in demand among 12% of industry employees), medical organizations (8%), retailers (6%) and the restaurant business (4%).

Usually, when evaluating offline communications, companies analyze only the speech of their employee using audio tabs, because when talking to several people, systems often cannot determine which of the interlocutors owns this or that replica. Yandex SpeechSense solves this problem neuronets YandexGPT with Speaker Labeling technology as well. They separate the replicas from each other in the dialogue and then distribute them to roles with high accuracy. Thus, a business can use audio recordings from any recording devices for voice analytics and see the results of analytics in the context of an employee and a client.