The artificial intelligence already began to transform the world of audiobooks with the ability to generate fluent voiceovers without using a human narrator, a novelty that concerns voice professionals. Tanya Eby He has been giving voice to texts full-time for twenty years.
In the last six months, his orders have halved and the activity of many of his colleagues has also slowed down. “It seems logical that artificial intelligence is affecting us“, says. “I think it is the plan for the future: replace workers and reduce costs“. There is no seal to identify the AI, but according to various professionals in the sector, thousands of audiobooks created from voice data banks are already circulating on the market.
(How Bard, Google’s artificial intelligence chat, works.)
Among the most advanced programs, deepzen offers a basic price that can cut the cost of creating an audiobook by up to a quarter compared to a classic project. This small London-based company works from a database that it created itself by recording the voices of various actors who were asked to express a variety of different emotions. “We have signed a license agreement with all the voices we use“, says Kamis Taylan, General Manager of DeepZen.
“We pay for the recordings (…) and we pay royalties every time we use the voice for a project“.”Things evolve so fast that there are many newcomers who do not have the same ethics” and use voices without paying their owners, warns Tanya Eby. “there is a gray area” that take advantage of many platforms, explains Kamis Taylan. “They take your voice, mine and five more, they create a new voice and they don’t pay anything, they say it doesn’t belong to anyone“.
(C-Pocket presents a solution for WhatsApp payments).
All companies contacted by AFP defend themselves against such practices. The Texan start-up speechki proposes, in addition to using his own recordings, use voices extracted from existing databasesrefers his CEO Dima Abramov. This second option requires the signing of a contract that includes the rights of use, he explains.
Asked about the issue, the five main publishers in the US market did not respond. But according to professionals interviewed by AFP, several names in the traditional publishing sector already use the so-called generative AI, that is, capable of creating, without human intervention, texts, images, videos or voices, from existing content.
(Expert warns of dangers of artificial intelligence).
“Professional storytelling has always been essential to listening at Audible and will continue to be so.“said a spokeswoman for the US audiobook giant, an Amazon subsidiary.”That being said, as technology improves, we envision a future where human interpretation and (AI) generated content can co-exist.“.
AFP
