Spotify is looking to use AI to translate podcasts into Spanish mimicking the person's style. This will use ChatGPT creator, OpenAI's, new voice generation technology, Whisper. French and German will follow later.
This is fine if done with the person's agreement and it seems Spotify is indeed doing this with the speaker's consent.
This raises the question of what happens when someone does it without consent. What protection does a speaker have? Does everyone have "personality rights" providing protection?
Certainly, use of photos could be protected via copyright or even as personal data under UK GDPR. Use of their name might be protectable via trade mark infringement or passing off.
But what about a person's speaking voice and style?
A person might be able to take action for passing off or unfair competition if it wasn't them actually speaking. But they would need to show the listeners were confused into thinking the translation was them speaking. This would likely apply only to well-known speakers. For example, King Charles III speaking fluent Spanish.
If a person's voice and speaking style were capable of being protected as a copyrighted work, then there might be some form of infringement of the right of reproduction or of the right to integrity. In reality, it would likely be the recording of the person's voice that would be protected by copyright as a sound recording. So, a translation could be an unauthorised adaptation or communication to the public. Or maybe their live voice and speaking style could be protected as a performer's right? Or is their voice capable of being personal data too?
Who will check the translation for accuracy? Current online translation tools give results with mixed success. Who is responsible if the translation is wrong? Or if a word is mispronounced to give an undesirable outcome?
AI is here to stay and we are only just lifting the lid on the possibilities - and all the tricky questions over legality that will go with it.
We are hosting a roundtable event on 25 January at our office in central London to discuss the risks. If you would like to attend or you need advice before then, contact me: firstname.lastname@example.org or +44 (0) 20 7611 2338.
According to Spotify, more often than not, "it is the voices of the speakers that lend as much weight to the stories as the narratives themselves."