Although still a rarity, a trademark can also protect a sign consisting of sounds, subject to the general conditions for trademark registration:

• the capacity to distinguish, in the course of trade, the goods or services of one natural or legal person from the goods or services of another natural or legal person, and
• the possibility of being represented in the Trademark Register in a manner that enables the competent authorities and the public to determine clearly and precisely the subject matter of the protection.

Sound Trademarks in Serbia – an Unprecedented Rarity

According to information from publicly available registers, only two trademarks in the form of sounds are currently in force as national trademarks in the Republic of Serbia: trademark reg. no. 51837, held by INTEL CORPORATION, and trademark reg. no. 74591, held by HYUNDAI MOTOR COMPANY. The remaining sound trademarks are international registrations designating the Republic of Serbia as one of the countries of protection. Both of these trademarks, as well as most of the international registrations mentioned, are represented in the traditional form of musical notation.

Given the small number of sound signs and the fact that no application for a sound sign has been filed with the Intellectual Property Office for almost a decade (the last application dates back to 2017), the Republic of Serbia is not, at this point, an adequate benchmark for examining the practical aspects of registering trademarks in this form. For that reason, the practical examples and trends should be examined primarily through the lens of international practice.

Changes in Serbia’s Trademark Law

 The currently applicable Trademark Law introduced one significant change six years ago.

Namely, the previously applicable law provided that trademark protection could be granted to signs composed of musical phrases represented in musical notation, which significantly limited the scope of protection available to sound signs. The valid Trademark Law sets out a more flexible rule, as it does not explicitly require musical notation as the mandatory form of representing the trademark, indicating that trademark protection may extend not only to musical signs that can be represented through musical notation, but also to other sound signs that meet the distinctiveness requirement.

Nevertheless, an open question remains as to how sound signs that cannot be represented through musical notation would satisfy the other condition set out by Serbian law – the possibility of representation in the Trademark Register – given that the current Trademark Register does not support audio recordings, unlike, for example, the EUIPO register or certain other offices, such as that of the United Kingdom.

International Practice: From Musical Phrases to Voice

International practice likewise cannot boast a large number of applications or registered trademarks in the form of sound signs. Nevertheless, this is expected to change in the future.

The most common sound signs are musical signs made up of musical sequences, such as several trademarks registered with EUIPO in March of this year by the rights holder GUERLAIN – Reg. No. 019272174, 019272128, 019272127, 019272129, 019272126 i 019272149.

However, beyond the basic function of a trademark – which is primarily to enable consumers to distinguish the goods/services of one market participant from the goods/services of other participants – in today’s legal ecosystem this basic function is changing and expanding.

The Italian singer Giusy Ferreri, known for the hits Non Ti Scordar Mai di Me and Roma – Bangkok (with Baby K), filed an application in April 2026 for a sound trademark consisting of the phrase „SONO GIUSY FERRERI“ (“I am Giusy Ferreri”) as spoken by her. The trademark in question is represented as an audio sign that can be listened to, rather than in the form of traditional musical notation, which in any case could not represent this phrase. The application was published on 8 May 2026, meaning that EUIPO has taken the position that the absolute grounds for trademark registration, including the distinctiveness of the sign, have been satisfied. If no opposition is filed by 10 August 2026, the trademark will be registered.

In this particular case, the focus shifts from the function of a trademark as a sign enabling consumers to distinguish goods and services on the market, toward a function of identity protection. This is, of course, not new, and this phenomenon is not solely a consequence of the development of generative artificial intelligence. Nevertheless, signs whose primary function is the protection of identity first raise questions as to their eligibility for trademark protection, and then, if registered, questions as to the scope of protection available to the rights holder in the event of unauthorized use of personal features, including replicas thereof. For now, it appears that no substantial changes have occurred, and that registering such trademarks will remain no easier than actually enforcing them in practice.

Although the application for this sound trademark successfully passed EUIPO examination and was published this past May, it is fair to ask whether this should have been the outcome, and whether such an application in fact meets all the absolute conditions for trademark registration.

It is fair to wonder whether the phrase will genuinely serve as an indicator of commercial origin, or whether consumers will instead perceive it as the very content of the product – Ferreri’s new musical single – and whether the average consumer would regard Ferreri’s voice as performing the function of a trademark for the goods and services applied for, which in this case cover goods in the respective classes.

In addition to the questionable distinctiveness of such a sign, one could also debate whether such trademarks run contrary to public policy, given that such applications unjustifiably expand the function of trademarks in an attempt to achieve goals that are not consistent with the purpose of the trademark protection system.

Scope of Protection: Is a Trademark an Adequate Means of Protection against Deepfake Content?

Even setting aside any doubt as to whether such a sign meets the conditions for registration, and even if we accept without objection EUIPO’s current decision on the justification, i.e., the fulfillment of the absolute conditions for registering this sound sign, including distinctiveness and public policy, the question of the practical scope of enforcement remains. Bearing in mind the list of goods and services for which it was filed, another question arises: can such trademarks actually protect against unauthorized digital replicas and deepfake content?

In principle, trademark infringement exists where the sign alleged to infringe the trademark is:

• identical to an earlier protected sign in relation to goods or services that are identical to the goods or services for which that trademark is registered, or
• identical to an earlier protected sign for a similar type of goods or services, or similar to an earlier protected sign for an identical or similar type of goods or services, if there is a likelihood that, because of that identity or similarity, confusion arises among the relevant public, which also includes the likelihood of association with that earlier protected sign, and
• it is used in the course of trade.

In this context, the question arises as to the actual scope of protection that such a trademark provides.

If, for example, someone were to use artificial intelligence to create a digital replica of Giusy Ferreri’s voice pronouncing the identical phrase “Sono Giusy Ferreri” and used that replica commercially for goods or services covered by the registration, for instance, in downloadable musical sound recordings falling under the registered class, it could be argued that this amounts to use of an identical sign for identical or similar goods, which would constitute a classic case of trademark infringement.

The situation becomes more complicated in other scenarios.

If a sound trademark were used for goods and services outside the registered classes, it would be necessary to assess whether there is a similarity of goods and services and a likelihood of confusion, rather than assuming automatic infringement.

If someone were to create a replica that merely imitates the timbre, tone and manner of speaking of Giusy Ferreri, but pronounces a completely different text, for example, reading the news or uttering some other slogan, then, in addition to the similarity or identity of goods and services, it would also have to be proven that an average user would associate such a replica with the singer, which is certainly no easy task.

If a third party were to commercially offer a voice-generation service similar to Giusy Ferreri’s voice from the registered trademark, for example, as an option in a text-to-speech tool without using the replica to pronounce the protected phrase itself, it would be difficult to speak of trademark infringement in the strict sense, although other legal remedies could potentially be invoked, such as personality rights, unfair competition, or personal data protection.

Likewise, if the digital replica is not used for commercial purposes, the question arises as to whether there is even use in the course of trade at all, and therefore whether any infringement of rights exists.

These examples show that a sound trademark consisting of a specific phrase provides relatively narrow protection, which is primarily triggered when a third party literally reproduces the protected sign for commercial purposes in relation to the goods and services applied for, while the broader phenomenon of voice deepfake replicas remains, for the most part, outside the reach of trademark law.

Between Registrability and the Actual Scope of Protection

Trademark law can indeed be a useful means of protecting the rights of a rights holder, including their personal features, such as their name, likeness, and even voice, but the system of protection it provides is far from complete. Different legal systems recognize different legal mechanisms such as copyright, performers’ rights, personality rights, and personal data protection which should be used together with trademark law.

Looking at the current trademark law in Serbia, it can be concluded that the previously applicable rule remains in force namely, that only musical signs capable of being represented in musical notation may be protected as trademarks, meaning that a spoken phrase, such as „SONO GIUSY FERRERI“, would not be eligible for trademark protection. The trend of expanding musical signs to sounds as a broader concept was also followed by the latest Trademark Law, which defines sound signs as eligible for trademark registration. However, as noted, the question of its full implementation, and of the possibility of representing such signs in the Trademark Register in the form of reproducible sound recordings, remains open.

This article is to be considered as exclusively informative, with no intention to provide legal advice. If you should need additional information, please contact us directly.