Become a member
2

MIN READ

May 1, 2024

HarperCollins UK partners with AI audio company

HarperCollins UK has announced a new partnership with AI audio research and technology company ElevenLabs to produce audiobooks in non-English languages. ElevenLabs offers text-to-speech AI voice generation in a number of voices and styles, and in 29 different languages, and has the ability to generate audiobooks in approximately an hour. 

According to The Bookseller, the partnership will produce audio versions of select backlist series books, which otherwise would not have been created. In the interim HarperCollins “will continue to devote time and resources to voice actor-led productions which are intrinsic to its current audiobook creation strategy.”

It is unclear whether this partnership extends to HarperCollins Australia, however, Australian authors who have books published by HarperCollins UK may be affected by this new arrangement.

We appreciate that publishers will increasingly look to incorporate AI into their workflows to save on costs and expand their offering but they must take their authors with them. The ASA expects that publishers obtain the consent of relevant authors to use an AI text-to-speech tool given that those authors would have granted audio rights at a time when synthetic voice was not prevalent or even imagined. Publishers realise the sensitivities around AI; they must know that for every author pleased that their AI-narrated audiobook will reach markets they otherwise wouldn’t reach, there are many more creators concerned about the broader trend of devaluing human contributions. The message we’re hearing from authors is a concern that this is the thin edge of the wedge.

In addition to concerns around creator solidarity, we have also heard from authors anxious to ensure the quality of text-to-speech technology is high and ensure a veto right if it is not. If a reader finds the listening experience poor, it can be the author who suffers when the reader dismisses the book as terrible or boring and decides not to buy anything else written by that author. We listened to ElevenLabs’ audio samples and even though they were impressive compared to previous benchmarks, synthetic voices cannot match the warmth, variation and performance of human voices, particularly over the duration of a book-length work. 

We also regard it as appropriate for agents/authors to renegotiate audio royalties for AI-produced audiobooks. ElevenLabs has said their technology “significantly reduces the costs and operational challenges associated with full-scale audiobook production.” Given that industry standard audio royalties were established factoring in high production costs and significant value-add, these changed circumstances should prompt a renegotiation of a fair royalty for these editions.

If we take this announcement as a sign of things to come, it’s now incumbent on every author to think carefully about granting audio rights and the conditions attached to that licence. You may be happy to grant your publisher broad audio rights, including for either synthetic or human narration in all languages, or you may want to impose conditions on the licence you grant to your publisher, for example by prohibiting AI-rendered translation, editing, and audio recordings without your prior consent. 

The ASA’s legal service Authors Legal is advising authors to include specific AI clauses in every publishing contract and you should expect your agent to discuss your preferences with you to do the same.

For advice on specific drafting, or if you are asked to sign a contract with an AI clause that you don’t understand, please seek guidance from Authors Legal. For all other advice, access our free Member Advice Service.

We will update our members with further guidance as more information about this deal becomes available.