Product reviews, deals and the latest tech news

Apple’s AI narrators are a long way from replacing human narrators in audiobooks

Apple has recently released its first batch of AI-narrated books, giving audiobook fans a glimpse into the future. Though this is an intriguing development with far-reaching consequences, the artificial narrators’ voices reassure listeners that their favourite human voices will be around for the foreseeable future.

To listen to audiobooks read by computers, just open the Apple Books app and do a search for “AI narration” to locate books read by computers. When you click on this, a list of free and paid romance and fiction novels narrated by Apple Books will appear.

Both the soprano, Madison, and the baritone, Jackson, AI voices available via Apple Books have American accents and can only be used to read aloud in English at the present time. You may hear a sample by clicking the “preview” button next to any narrated book on Apple Books.

Both of Apple’s AI voices have a robotic, unnatural tone at the moment. You won’t be fooled into thinking it’s prominent narrators like Stephen Fry or Julia Whelan providing the voices. AI narrators are unquestionably on the fast road to our ears, but the uncanny valley remains a difficult challenge to overcome.

Apple claims that the creation of its artificial intelligence voices was motivated by a desire to make audiobooks “more accessible to everybody.” In addition, they facilitate entry into the multibillion-dollar audiobook sector for technological behemoths. Also, Apple’s new Books feature is only the beginning of a heated competition with other audiobook providers like Amazon and Spotify.

Distressed expressions

We should expect to see AI narrators in novels mostly from smaller, independent publishers for the time being. As digital narration creates a new audience for publishers and writers who couldn’t otherwise afford to make the transition from print to audio, this might lead to a dramatic increase in the number of audiobooks accessible to listeners across all platforms.

Nonetheless, resistance from both major publishers and voice performers may hinder the development of robo-narrators. Although there are workarounds in the accessibility settings, Amazon’s Kindle e-readers no longer have text-to-speech capabilities. Legally speaking, audiobooks are treated as their own works of art, therefore concerns about copyright played a role in this decision.

Audible, an Amazon subsidiary, has written extensively on the best narrators for various genres of books and how publishers go about selecting them. When it comes to audiobooks, the company stresses how crucial it is that the narrator’s voice be appropriate for the story’s style and mood in a blog post. This is harder to do if, like Apple Books, you only have two voices.

And this is certainly where AI voice actors need to do some intense vocal training, and perhaps also take some night classes in emotional mirroring, as Audible puts it: “experienced voice actors are able to differentiate pretty easily between vocal characteristics by playing with pitch, intonation, volume, and accents.”

However, it’s evident that the new Apple Books feature is simply the beginning of a massive expansion of speech technology powered by artificial intelligence. And the true game-changer for audiobooks might come when AI can effectively emulate a renowned voice actor rather than robotically read a script, a step that may not be too far off based on recent demonstrations by companies like Amazon.

Faked-up hegemony

Since AI-powered speech technology is also rapidly progressing, deepfakes are not limited to shockingly realistic movies of a synthetic Morgan Freeman or Tom Cruise.

Web apps like Uberduck allow users to generate speech in the voices of fictional characters like presidents or cartoons, and just last year Amazon demonstrated a slightly unsettling new Alexa skill that could read The Wizard of Oz to a child in the voice of the child’s grandmother.

In the future, AI voices will be even more convincing because to developments by companies like Google Wavenet. Rather than technical limitations, issues of law and ethics may now stand in the way of audiobooks’ wider dissemination. Apple Books’ synthetic narrators, on the other hand, are the sound of a synthetic future in which the voices of renowned people will be copyrighted and licenced.

The data in Apple Books shows that AI voices are presently most appropriate for non-fiction and factual literature rather than emotional narrative. Novels, like movies, have a ways to go before digital performers can successfully pull at our heartstrings without a synthetic accent or flat note. Even so, a deluge of AI-related audiobooks is on the way.