Through artificial intelligence, Spotify is expanding the markets available to its podcasters in other languages.
The company on Monday unveiled a test technology for translating podcasts from one language to another while preserving the podcaster’s voice, called Voice Translation.
With the help of OpenAI’s speech creation technology, Spotify’s new translation tool can mimic the voice traits of a speaker to make translations sound more authentic.
Selected podcasts by Dax Shepard, Monica Padman, Lex Fridman, Bill Simmons, and Steven Bartlett will be included in the pilot program and will be translated into Spanish, French, and German.
The upcoming episodes of “The Rewatchables” from The Ringer, Dax Shepard’s “eff won with DRS,” and Trevor Noah’s upcoming original podcast will all be translated by Spotify.
Voice Translation, according to Spotify Vice President of Personalization Ziad Sultan, “gives listeners around the world the power to discover and be inspired by new podcasters in a more authentic way than ever before by matching the creator’s own voice.”
“We believe that a thoughtful approach to AI can help build deeper connections between listeners and creators, a key component of Spotify’s mission to unlock the potential of human creativity,” he continued.
Benefits for Spotify and Podcasters
The new translation tool might be useful for Spotify and podcasters alike. The Spotify suggestion, according to Greg Sterling, co-founder of the news, opinion, and analysis website Near Media, “could broaden the audience reach of these podcasts to new audiences and nations.”
He told TechNewsWorld that by extending the podcaster’s audience, “Spotify and the podcaster could both benefit from this.”
According to Rowan Curran, an analyst with Cambridge, Massachusetts-based Forrester Research, English podcasts that have been translated into Mandarin and Hindi would have access to some pretty sizable markets that they wouldn’t otherwise.
He told TechNewsWorld that this “represents a democratization of language AI capabilities.” That is consistent with the prior couple of years, when a wide range of people were able to use these incredibly powerful functionalities.
According to Rob Enderle, president and lead analyst of the Enderle Group, an advisory services company in Bend, Oregon, podcasters will not only be expanding their audiences but also their bank accounts because the more potential customers their podcasts attract, the more money they may be able to make.
For Spotify, the same holds true. According to him, “Each performer can generate more income; high performers will make the company much more money,” he told TechNewsWorld.
The Pressure to make Investments Profitable
The co-founder and CEO of Pleasanton, California-based generative AI company Gleen, Ashu Dubey, concurred that the translation tool might boost Spotify’s earnings.
According to him, this technology might help Spotify sell more subscriptions in France or Japan, for example, if there is a popular podcast that is only available in English. He made this statement to TechNewsWorld.
According to Todd Cochrane, CEO of Blubrry Podcasting, a podcast hosting and distribution provider in Traverse City, Michigan, Spotify actually needs to sell more subscriptions.
As they are under intense pressure to make their billion-dollar investments make up for the money they have lost, he told TechNewsWorld, “They need bigger numbers of listeners to monetize against.”
Recent high-profile transactions struck by Spotify include a multi-year, US$200 million exclusive agreement with podcaster Joe Rogan, a $196 million purchase of the sports and pop culture website Ringer, and a $56 million investment in the Parcast production firm, which is best known for its true crime podcasts.
While Spotify now leads the pack in terms of its translation tool, this advantage may quickly fade. Although Spotify is the first major creator platform to implement this, Curran warned that it won’t be long before similar technology appears on other sites like YouTube.
Technology that Might Be Dangerous
Despite the advantages of Spotify’s new translation feature, the technology behind it also has a negative aspect.
According to Sterling, the technology “can be quite dangerous and potentially exploitative.” It has already been employed in fraud and con schemes. Additionally, recordings of audiobooks already use celebrity voice clones without authorization.
It must be used carefully and always with the subject’s consent, he continued. “However, the disparity of power between platforms and the users who utilize them may prevent the development of voice AI use cases that are fair. There must be clear, moral standards in place.
“This is one of the problems with the ongoing actor strike. Do the studios have the right to use an actor’s voice and image without their consent for all time? Added he.
Dubey made the observation that the translation tool might be vulnerable to hallucinations, the bane of AI applications.
“This could happen if the podcaster were to use a phrase that didn’t really have an equivalent phrase in the language being translated,” he said.
“For instance, the German word schadenfreude doesn’t really have a strict translation in most languages, so an AI that is relying only on a large language model could end up hallucinating the translation and inserting words into the podcasters’ mouths,” he explained.
The Success Factor Is Execution
Podcasters might face legal issues as a result of translations.
Alyssa J. Devine, CEO and founder of Purple Fox Legal, a Nashville, Tennessee-based law firm with a focus on intellectual property law for entrepreneurs and creatives, warned that “if the AI technology fails to provide an accurate translation of a podcast creator’s content, the podcast creator could face legal consequences, such as defamation or FTC violations.”
It is not unusual for a plaintiff in one country to get a judgment against a defendant in another country, she told TechNewsWorld. “The appropriate jurisdiction and venue for such claims would depend on the facts of a specific situation.”
Voice Translation will succeed through execution, according to Cochrane.
“If Spotify does not execute this well, it could do the opposite and hurt all podcast content across the platform and turn those non-English native listeners off to the content,” the author said. “If it sounds artificial and lacks inflection, there is a real risk.”
It can be difficult to translate podcasts, according to Mark N. Vena, president and principal analyst of SmartTech Research in San Jose, California, and a podcaster.
“When you translate things into different languages, everything said in one language can’t be cleanly translated into another,” he claimed.
It will be a problem if the translation’s accuracy isn’t very good, he continued. Cleaning up some of a podcast’s artifacts, such as odd gaps and “ums” and “ahs,” will also be a challenge.
He declared, “I’m really suspicious of how effective this will be.