In the voice social network Clubhouse appeared Assistant Tinkoff named Oleg. This is the first case of integrating voice assistant and synthesis technologies and speech recognition on this platform.
As it turned out to be an assistant, thanks to its recognition skills and synthesizing speech online, it will be able to function in the social network as a full-fledged Internet user providing room moderators for additional contact tools with subscribers and discussion modeling. The voice assistant will be able to attend the service room and listen to speakers at the same time decrypting texts and leading live broadcasts in their Telegram channel "Oleg in Clabukhause". He will be able to act as a moderator to voice the words of the speakers to remind of regulations and timing and perform a lot of other actions reports Live24.RU
Assistant March 11 has already spent the first broadcast with text decoding from the Tinkoff Room. During the event with journalists and investors, the Top Manager of Oliver Hughes was broadcast.
The discussion in the room was devoted to the topic of financial results and record profits of the group
activities for 2020. The director of Technology II Tinkoff Pavel Kalaidin said that at this stage the team experiments with various scenarios in Clubhouse which could be useful for developers and interesting for students. Testing the conversion tool of the audio signal to the text in your own
It turned out to be successful.
Oleg's voice assistant is also useful for the creators of the rooms and discussion moderators when there is no opportunity to ask a question with a voice. For such cases, developers have prepared an interface that allows you to make questions through chat with an assistant in Telegraph. Oleg is able to not only perfectly read the texts but also to preserve the anonymity of the authors. At this stage, work on improving the assistant continues. Among other things, the possibilities of totaling information and text cleansing so that the transcription is more readable.
The presented voice assistant is based on Tinkoff Voicekit's speech technologies. These are deep neural network models developed in the framework of the AI FIRST project. This tool was used as well as to create an assistant named Oleg who is the world's first financial voice assistant to develop Tinkoff. Tinkoff VoiceKit technology can be used to create robots to automate work in call centers for quick recording audiobooks as well as editing and voicing video.
These models will help in developing their own
and programs for people with disabilities. In addition, these tools can be used for search engine optimization including to search for audio and video transcription of any public sound recordings and constructing a speech analytics system for transcribed sources.
Pavel Kalaidin stressed that in the group ready for joint experiments with clubhouse communities because they hope that Oleg will become a good assistant for listeners and content-makers. For cooperation, please contact [email protected]. You can leave the Tinkoff Voicekit application by reference
.