Yandex.Cloud introduced the Yandex SpeechKit Brand Voice technology, with the help of which you can synthesize a new unique and indistinguishable from a human voice for your company.
The solution will allow companies to stand out from competitors and spammers, increase customer loyalty, and open up new opportunities for advertising campaigns.
The technology is developed on the basis of machine learning models that are trained using recordings of a real person’s speech and produce a complete digital copy of the voice. It also allows you to automate communications with customers several times faster than other solutions on the market.
Since July 2021, SpeechKit Brand Voice has been presented in Full and Adaptive versions and is available on the Yandex.Cloud website.
With the Adaptive version, a unique voice is created in just one month. This solution is suitable for the most demanded business tasks:
- Call routing
- Telemarketing/Automated Voice Calls
- Confirmation of orders, appointments, reminders
- Polls and NPS measurement
The Adaptive version is based on the synthesis of variables – the development of Yandex.Cloud has no analogs in the world. It is unique in that it creates a digital copy of the voice of a specific person according to predefined templates in just a few hours, and then personalizes the speech: it adds the desired appeal, the details of a specific order, the amount, and so on.
At the same time, the synthesized voice sounds as natural as if the phrase was pronounced by a real person.
The Full version is designed to solve complex problems, such as:
- Voice assistance in mobile applications
- Handling incoming calls to your number, IVR
- Dubbing books and any text content.
The implementation time for an interactive robot is approximately six months, and the cost of developing a model is determined individually, depending on the complexity of the client’s tasks.
The Full version is a complete digital copy of a real person’s voice without restrictions. To “train” such an assistant, about 50 hours of speech recording are required, after which the robot will be able to pronounce any text in a unique voice.
Source: Yandex Cloud Blog