In the last few hours, the news regarding Prime Minister Giorgia Meloni who has filed a request to register her voice as a sound trademark to combat audio deepfakes with the EUIPO, the European Union Intellectual Property Office, has aroused some curiosity (for the record, at the current state of affairs the Prime Minister’s application is still being evaluated by the office). It is not the first time this has happened since in the past other celebrities – such as the singer Giusy Ferreri and the voice actor Luca Ward – have done something similar to combat the phenomenon of voice cloning done with AI. Anyone can apply to the EUIPO and do the same thing even if they are not in the celebrity sphere, but to combat audio deepfakes there are many other things that we can all do to mitigate these new risks attributable to artificial intelligence. Registering your voice as a sound trademark is a legal protection, not a technical protection from deepfakes.
What is a sound trademark, how to register the voice and the EUIPO requirements
Recording your voice and transforming it into a sound brand can represent one of the ways to evaluate when you want to protect a distinctive element linked to your identity. But be careful: this does not mean that given the proliferation of AI tools we should all en masse record our voices to defend ourselves from audio deepfakes. The EUIPO procedure, the European Union Intellectual Property Office, concerns the registration of a sound trademark and does not represent a technical system that prevents artificial intelligence from cloning a voice.
A sound trademark is, by definition, a trademark consisting exclusively of a sound or a combination of sounds. If you want to register a sound as a trademark and make the procedure valid within the European Union, you must submit an application to the EUIPO.
According to the office’s guidelines, the file to be uploaded to the platform must meet certain specific requirements:
- It must necessarily be in MP3 format;
- It cannot exceed 2 MB.
- Must have a sampling rate of 8 kHz for voice only and a bit depth of 8 or 16;
- A recording that reproduces the audio in a loop, therefore automatically and continuously, is not permitted.
The audio file can be transmitted exclusively through the e-filing system, following the online procedure provided by the office for submitting applications. The procedure can be started by going to this page, clicking on the yellow Start application button and registering on the platform using the appropriate form. By following the procedure that appears on the screen, you will then be guided in forwarding the request.
Please note that the procedure is not free: you have to pay the basic fee within one month of receiving the application from the EUIPO. The Office proceeds with the examination of the application and with the related communications only after payment.
The basic fee for an online application is 850 euros, as specified on this information page. Given the cost to be incurred, it is clear that resorting to the EUIPO makes sense if your voice truly represents a distinctive element that you wish to legally protect. If you are simply afraid of audio deepfakes and want to protect yourself from them, submitting an application takes your time: as already mentioned, filing the sound trademark is a form of protection, but it does not prevent the AI from cloning the voice, even if it has been filed with the office.
Why do it: reduce the risks associated with voice cloning
To reduce the risks associated with voice cloning, it is much more important to adopt some precautions in managing your voice, which can really make a difference. You must avoid, for example, unnecessarily disseminating particularly long and easily reusable personal voice recordings and pay attention to the audio content you make publicly available. This could mean setting your social profiles to private, not sending voice messages to strangers who contact you on messaging apps and paying attention to silent calls, avoiding starting to talk if the person on the other end hasn’t introduced themselves.
