Voices are stolen from announcers – Newspaper Kommersant No. 15 (7460) of 01/27/2023
[ad_1]
The Union of Speakers is asking the State Duma to establish a special regulation for the synthesis of the human voice using artificial intelligence (AI) technologies. Announcers talk about instances of their votes being stolen and fear a decline in income amid the spread of technology. It is proposed, in particular, to introduce licenses for the use of a certain voice. Speech synthesis companies say they care about the legitimacy of the technology. But lawyers confirm that the voice is in no way protected from copying.
“Kommersant” got acquainted with a copy of the letter from the Union of Speakers of Russia to the head of the State Duma Committee on Information Policy, Alexander Khinshtein, dated January 26. The announcers propose to develop regulation of voice synthesis with the help of AI. The union confirmed the authenticity of the letter, Mr. Khinshtein told Kommersant that he had not received it yet.
We are talking about technologies that copy a person’s voice based on a sound recording and allow it to be further used in voice assistants, online synthesizers, etc. In the future, it will become cheaper to get a synthesized speech than to record an announcer, and many voice artists will lose their jobs, the authors believe. letters. The announcers in the letter talk about the facts of “stealing” the voice with the help of AI: “There are no norms, no practice, no understanding of how to write a claim. Customers refer to the contract, which contains clauses about the alienation of rights, the right to processing and the right to transfer to third parties.
Members of the union propose to consolidate the ban on the alienation of the rights of performers when using voice synthesis and switch to license agreements “with a clearly defined period of validity”, as in the field of protection of trademark rights. It is also proposed to fix the rule that the recording of a synthesized voice can only be used in a specific project.
AI tools are beginning to be actively introduced for sounding audio books, Kommersant reported on January 23: market participants expect that this will reduce the cost and speed up the process of their production. “Publishing houses need good voice acting at an affordable price, announcers need to preserve the profession that many have given years to,” says Nikolay Boronin, head of audio projects at Alpina Digital. The owner or developer of AI must have a contract to use the voice, but no one is safe from outright piracy, he admits.
The publishing group “Eksmo-AST” believes that the fears of the announcers “are not fully justified.” The Yandex service SpeechKit, with which the group cooperates, enters into agreements with the announcer, on the basis of whose voice speech is synthesized, the group’s press service says. In addition, without the personal participation of a specialist whose voice is selected according to special parameters, it is impossible to achieve the quality of automatic voicing, they add in the group.
The head of VS Robotics (engaged in solutions based on speech technologies) Dmitry Teplitsky said that in theory it is possible to synthesize the speech of a certain person without many hours of recording: “We have experience in creating the voices of famous personalities based on recordings of performances, excerpts from films, public speaking. It is important here that the records are as clean as possible and accompanied by a text transcript.” But to create a new voice for TTS (text-to-speech, text-to-speech), the company still selects speakers and records their live voice for five hours or more. STC CEO Dmitry Dyrmovsky notes that leading companies in the field of speech technologies pay attention to the legitimacy of its use “and, in general, the ethics of AI.”
The voice cannot be protected from “copying”: if the imitation itself is performed without criminal intent, then there is not even any liability for such use, says Yury Fedyukin, managing partner of Enterprise Legal Solutions: “You can understand the speakers, but from the point of view of the law, the implementation of their initiative is impossible simply due to the fact that it is not the voice that is protected by law, but the phonogram created with its use. Making changes to Russian legislation that could make this possible does not make sense, Yuri Fedyukin believes, since then the voice recordings of the announcers will simply be processed in other countries.
[ad_2]
Source link
