Lines (Book Service)

History

2024: Transition under the management of MTS Media

On May 16, 2024, MTS, a digital ecosystem, informed TAdviser that MTS Media holding had completed the consolidation of the KION online cinema, the String book service, MTS Music streaming and MTS Live and MTS Label directions into a single vertical. Read more here.

2023: The start of using the neural network to voice books

On January 23, 2023, it became known that Russian book services began to use neural networks to voice books. In particular, the portal "Strings" (owned by MTS) is engaged in such a project, which by the end of 2023 is going to voice more than 10 thousand works using artificial intelligence MTS AI.

As told To the businessman"" in MTS, by January 23, 2023, 600 works placed on the "Lines" site are in work. Books are recorded using the Audiogram platform, which uses neural networks, machine learning and natural language processing technologies (). NLP

Russian book services began to use neural networks to voice books

According to MTS, by January 2023, more than 90% of books on the Russian market do not have an audio version. Voicing them in traditional ways would take years. Using AI helps speed up and reduce the cost of the process. Voicing the book through a neural network begins with an initial preparation that takes 30-60 minutes. It takes several days to record the announcer in the studio, taking into account sound processing and editing. Also, the work of the announcer costs 50 thousand rubles per book, and the cost of voicing one book using AI costs 400-700 rubles.

The leader of the audiobook market - LitRes Group of Companies (LitRes, MyBook, Livelib, etc.) also develops audiobook voicing using AI, Evgeny Selivanov, director of the group's content development department, told the newspaper. According to him, the work uses SpeechKit speech recognition and synthesis technology, on the basis of which the voice assistant Alice operates. Selivanov noted that the algorithm allows you to produce several times more books per period than a distributed team of readers. Unlike standard robotic voice acting, he said, AI remembers pauses marked by the editor, arranging them on its own, which avoids distortion.^[1]

Notes