Image – Getty: Sonal Garg
By Porter Anderson, Editor-in-Chief | @Porter_Anderson
See also:
The Publishers Association on the UK Government’s AI Consultation
‘To Promote Consistency Across the Industry’
As many controversies and concerns around artificial intelligence and book publishing have made their way through 2024, one aspect of the discussion has involved how the industry should label audiobooks that are voiced by AI systems, rather than by human readers.
Today (December 19), the United States-based Audio Publishers Association and the United Kingdom’s Audio Publishers Group—part of the UK’s Publishers Association (PA)— have together issued a kind of holiday gift to the industry: a concise guide for publishers and retailers to consider around issues of responsible consumer labeling of audiobooks on which the readings are produced with AI technology.
There’s an interesting caveat in the document released today, saying “These guidelines make no judgments on the use of AI narration in the industry.”
That’s a reflection of how sensitive many in the industry are to AI-related topics, and this is, of course, particularly true in the audio realm in which human readers may object to machines performing services that humans would be paid to provide. On the other hand, of course, automated readings can make it affordable for publishers to consider audio-format releases of their backlists or other titles they might not have the budget to produce using human readers.
“By working together to create thoughtful guidelines, we help to ensure transparency, protect consumer trust, and recognize technological changes, all while maintaining the integrity of the audiobook experience.”Michele Cobb, Audio Publishers Association
In another instance, the document reads, “It should be stressed that these guidelines are best-practice recommendations only. They are not requirements. However, they are the considered conclusions that emerged from cross-industry discussion.”
Jon Watt of Bonnier Books UK has been at work on this with the PA’s Audio Publishers Group for many months, and was unfortunately unable because of a transit delay to join us as planned at Frankfurter Buchmesse to brief us on this work in our Publishing Perspectives Forum in October.
Watt is joined by the Audio Publishers Association’s executive director Michele Cobb in developing and introducing these guidelines and the rationale for them.
“At present,” Cobb and Watt write, “various terms for different kinds of AI narration are being used interchangeably in the marketplace, which leads to confusion among publishers, retailers, and consumers. The new guidelines have been produced in consultation with publishers, retailers, distributors, and metadata providers from the United States, Canada, the United Kingdom, and Europe.
“They aim to promote consistency across the industry to give consumers clear information and allow them to make informed choices when purchasing audiobooks.”
Watt and Cobb go on to say, “The focus in selecting terms was that they would be easy for publishers and retailers to use and interpret, and for consumers to understand.”
Points From the Guidelines
Our international readership will want to read the full document as released today. It includes title metadata instructions for publishers and retailers. The new guidelines are designed to address two kinds of AI voicing.
“AI Voice” is defined as:
- “An AI-based synthesized voice that has been generated using samples from a large group of unidentified speakers. An example of this is Google’s “auto narration” voices, such as ‘Archie’ or ‘Mary.’” An example offered in the guidelines:
Pride and Prejudice
Author: Jane Austen
Narrated by: Amir (AI Voice)
Duration: 11 hours
Pub date: 01-01-29
“Authorized Voice Replica” (AVR) is defined as:
- “An AI-based voice that has been generated using authorized/licensed samples from a specific human voice and seeks to replicate that voice.
- “An example of this would be a publisher working with a deceased author’s estate to create an authorized voice replica based on archive samples of the author’s voice.”
AVR guidelines include a note pointing out that replication of specific human voices is sometimes referred to as cloning. “For the purposes of these guidelines,” the document clarifies, ‘cloning’ refers to unauthorized replication where a human has not given permission for their voice to be replicated.’”
What’s more, the document points out, “Most AI voices have identifying names, in some cases a single given name. These guidelines encourage the use of this name as the ‘narrator’ or ‘reader’ in the title metadata, allowing consumers to easily identify and select voices, as they do with standard human narrators.”
An example of this in the guidelines:
The Hound of the Baskervilles
Author: Arthur Conan Doyle
Narrated by: Jane Smith (Authorized Voice Replica)
Duration: 6 hours
Pub date: 01-01-29
In this section dealing with the “authorized voice replica,” it’s pointed out that the “real name” can also be given an identifier such as an ISNI (see https://isni.org), to ensure there’s no ambiguity with other people of a similar name.
Comments from Cobb and Watt
Michele Cobb
In remarking on the release of these cleanly conveyed new guidelines, Cobb at the Audio Publishers Association says, “By working together to create thoughtful guidelines, we help to ensure transparency, protect consumer trust, and recognize technological changes, all while maintaining the integrity of the audiobook experience.”
And Watt, chair of the UK Publishers Association’s Audio Publishers Group, says, “With the proliferation of AI-narrated audiobooks, it’s vital that our industry has an international language that allows publishers, retailers, and consumers to distinguish between different narration types.
Jon Watt
“It is only by giving consumers the chance to make informed choices around AI-narrated products that we as an industry can maintain consumer confidence in our products.
“To bring the broadest benefit, it’s our hope that publishers and retailers will adopt these naming practices and convey them to consumers.”
The Question of the Term ‘Narration’
The reference at one point in this labeling-guidelines document to “narrator or reader” may be of interest to those who have realized that narrator might not be as accurate a term as reader.
“To bring the broadest benefit, it’s our hope that publishers and retailers will adopt these naming practices and convey them to consumers.“Jon Watt, Bonnier Books UK
Narration is “the act or process of telling the particulars of an act, occurrence, or course of events,” according to Merriam-Webster’s first definition—in other words, talking about something, not reading its text aloud. To hear a bit of true narration, listen to John Hurt’s opening lines in the 2006 DreamWorks film adapatation of Patrick Süskind’s Perfume: The Story of a Murderer from director-composer Tom Tykwer. You can hear this in the film’s first trailer here.
A book’s text may actually include narration, of course, but even so, the oral delivery in an audiobook of that narration is being read or voiced.
As the audiobook industry matures—and such useful exercises as the development of this labeling protocol are developed—a more accurate term for the oral delivery of a book’s text may gradually prevail.
More from Publishing Perspectives on audiobooks is here, more on the Audio Publishers Association is here, more on the UK’s Publishers Association is here, more on industry statistics is here, more on the United States’ market is here, and more on the United Kingdom’s market is here. More on digital publishing is here.
About the Author
