Protection of Voice Actors' Rights and Interests in the Context of AI

Speech Synthesis Technology

Jiahan He

Law School, Hubei University of Economics, Wuhan, Hubei, 430205, China

Keywords: Dubbing Industry, AI Speech Technology, Voices Infringement, Vioces Protection.

Abstract: The development of generative artificial intelligence (AIGC) technology is advancing rapidly, accelerating

the transformation of digital audio content production models while also triggering a series of legal risks,

among which voice infringement issues are particularly prominent. Giv-en the unique industry characteristics

of the dubbing field in China, it is imperative to regulate AIGC technology through legal means. Starting from

the first judicial dispute over AI-generated voice infringement in China, this article attempts to deconstruct

the phenomenon of voice infringement, finding that it is closely related to the transformation needs of the

dubbing industry and the weak protection of the law. Based on this, it proposes relevant protection measures

such as independent legislation, improving voice authorization, and establishing a voice evaluation

mechanism, with the aim of exploring protection paths for dubbing actors' rights in the digital age and

achieving a mutually beneficial interaction be-tween AIGC tech-nology and the dubbing industry.

1 INTRODUCTION

With the continuous maturity of Artificial

Intelligence Generated Content (AIGC) technology

in China, human-computer interaction scenarios have

become increasingly rich. Technologies such as

voiceprint recognition and speech synthesis have also

found diverse application scenar-ios in daily life.

However, as a representative personality right and an

important commercial resource, the risk of

infringement of voice's biological information is

correspondingly increas-ing. Against the backdrop of

infringement chaos caused by technology abuse,

China has not yet issued laws with artificial

intelligence as the main subject, which makes it more

difficult to identify the responsibility and divide the

rights and responsibilities of voice infringement in ju-

dicial practice.

This study aims to, within the existing judicial

framework, through the rational transfer and ap-

plication of similar systems, conduct an in-depth

analysis of the issues related to the identifica-tion and

protection of sound rights and interests from three

dimensions: promoting the inde-pendent legislation

of voice rights, improving the sound authorization

system, and constructing a voice evaluation

mechanism.

2 THE URGENCY OF

PROTECTING AGAINST THE

PHENOMENON OF AI VOICE

INFRINGEMENT

2.1 The Particularity of the Dubbing

Industry and the Need for

Transformation

Sound is a unique acoustic phenomenon produced by

the vibration of a person's vocal cords. Due to the

differences in the structure of each person's vocal

cords and oral cavity, everyone's voice has its own

uniqueness (Schierholz, 2019). Voice actors can

achieve the performance effect of playing multiple

roles with one voice by changing various aspects such

as the pitch and sound pressure of their voices. As can

be seen from the judicial precedents of the first AI-

generated voice personality right infringement case in

China, the protection of voice rights and interests

today not only has no systematic and professional

legal support, but also the variable sound color and

tone line creates a protection dilemma for the

legislation of the voice itself. This undoubtedly

further impedes the development of the industry.

Also, allowing the proliferation of AI-infringing

He, J.

Protection of Voice Actors’ Rights and Interests in the Context of AI Speech Synthesis Technology.

DOI: 10.5220/0014380600004859

Paper published under CC license (CC BY-NC-ND 4.0)

In Proceedings of the 1st International Conference on Politics, Law, and Social Science (ICPLSS 2025), pages 347-352

ISBN: 978-989-758-785-6

347

products, with a large amount of unauthorized AI

voiceover content flooding the market, will reduce

the commercial trust of the audience in dubbing

works, affect the reputation of the entire industry, and

further reduce the demand for professional voice

actors, thus impeding the sound development of the

industry.

The transition of voice actors from behind the

scenes to the spotlight is a symbol of occupational

standardization and a necessary step for the

transformation and development of the industry. This

enables the listener to better establish the connection

between a particular voice line and the sound source,

realizing the “many-to-one” correlation effect, further

laying a feasible foundation for the recognizability of

the voice, and alleviating the difficulty of applying

the extremity of variable voices. In addition, dubbing

is a performance activity that integrates "emotion,

intonation, and breath." Hastily divested of human

understanding and generalized digitization of the

initiative itself has limitations. Because the volume of

AI-generated sound is relatively constant, it is

difficult for the listener to perceive the three-

dimensional auditory space, and thus sample the

immersive experience brought by dubbing

performance. When the portability of voices

technology is emphasized far more than the pursuit of

aesthetics, the phenomenon of bad money driving out

good will inevitably occur, squeezing the living space

of practitioners and raising the threshold of entry for

newcomers. In the long run, this will not only reduce

the enthusiasm of the dubbing ecosystem, but will

also be detrimental to the transformation and

upgrading of the dubbing industry.

2.2 The Weak Protection of Voices by

Existing Laws Abstract Frame

Compared with representative portraits, auditory

sounds have long been on the marginal position in

terms of legislative protection. Although previously,

relevant provisions regarding sound rights and

interests have been made in different forms in legal

norms such as the Trademark Law and the Anti-

Unfair Competition Law, on the whole, these

provisions tend to protect the economic rights and

interests generated by sounds, with relatively little

protection for the sounds themselves. With the

increasing frequency of voice infringement cases

recently, and the judicial practice facing the

embarrassing situation of having no laws to rely on,

there is an urgent and realistic need for legislation on

the right to voices.

Throughout the world, the United States protects

voices' interests through a dual legislative model of

the right to privacy and the right of publicity.

However, Liming Wang on the protection of the

migration of the model of the application of the denial

of the right of publicity as the United States as the

original concept of the rule of law, from the concept

of the creation of the object it protects are not

applicable to the Chinese system. Article 9 of the

French Civil Code stipulates that, as one of the

personality characteristics, when the voices meet a

certain degree of subject recognizability, they can be

protected by an independent right to voices.

According to Article 36 of the Civil Code of Quebec,

Canada, in this region, names, portraits, sounds, etc.,

all fall within the scope of the extended rights and

interests of the right to privacy, and they are protected

by safeguarding the right to privacy in judicial

practice. Germany adopts a criminal legislation

model, protecting the voices as an independent right

of personality through criminal law (Wang, 2024;

Chen,1981).

In summary, in addition to the Canadian province

of Quebec and other geographical areas, most

countries or regions of the law to a certain extent,

recognized the voice of the status of independent

personality rights, effectively demonstrating the

inevitable development trend of the legalization of

voice rights and interests on a global scale.

Looking back on the research by Chinese

scholars, Guodong Xu believes that the “portrait and

voice rights" should be combined to create the same

legislative protection for sound and similar portraits.

However, the author of this paper holds the view that

the establishment through combination implies the

acknowledgement that the legal interests of the two

are different and that the existing laws are imperfect.

Given the irreversible development of AIGC in

today's era, this view precisely corroborates the

theory of independent legislation for voices proposed

by Lixin Yang. Liming Wang once advocated that the

voice is not an independent personality right, since

the promulgation of the Civil Code, changed his view

that the voice is a special legal personality interest,

not a specific personality right; and scholars

represented by Lixin Yang believe that the right to the

voice is a natural person to independently dominate

their own voice interests, decide to use and dispose of

their own voice of the specific personality right, the

right to the voice should be independent (Xu, 2004;

Wang, 2018; Yang, Yuan, 2005). Article 1023 (2) of

the Civil Code for the first time on the protection of

voice “reference to the application of” portrait rights

of the quasi-legislative technology. But the rights and

ICPLSS 2025 - International Conference on Politics, Law, and Social Science

348

interests of the voice and the protection of the right to

portrait in the infringement of the elements and forms

of different, and the voice does not need to be a carrier

to break through the limitations of the right to portrait

the need for carriers, which gives rise to a variety of

insufficient to be “reference to the application” of the

right to voice. The differences are not sufficient to be

covered by the “application by reference”, so people

need a strong relevance of the legislative guarantee.

3 APPROACHES TO

SAFEGUARD THE RIGHTS

AND INTERESTS OF VOICES

Indeed, the widespread application of AI in the field

of voices will indeed squeeze the survival space of

dubbing practitioners. However, in the context of

technological empowerment, people cannot be afraid

of the pain of change in the industry to stand still. The

key contradiction is that the relations of production

represented by the law have not been able to adapt to

the productivity changes brought about by artificial

intelligence promptly, and stopping it at this time is

tantamount to holding on to the past. It is not wise to

curb the development of produc-tivity, but it is

necessary to proactively introduce relevant policies

and measures. The following three solutions are

proposed in order to realize the protection of the

rights and interests of voic-es..

3.1 Promote the Independent

Legislation of the Right to Voice

3.1.1 What Is the Legal Attribute of the

Right to Voice?

The importance of voice legislation has already been

discussed, but since China has not yet introduced an

enforceable statutory law, it needs to be explained

from the perspective of feasibility. Many countries

define image rights as covering the right of

individuals to prohibit the unauthorized use of their

names, portraits and voices. However, at this stage,

the EU lacks a unified and coordinated image rights

framework, and there are significant differences and

fragmentation in the image rights systems of

countries around the world. Such as some legislative

initiatives in Russia, which analogize the protection

of voice to the protection of the human image.

However, the legal definition and protection scope of

voices remain unclear. This makes it difficult for

voice actors to find clear and powerful legal bases to

safeguard their own rights and interests when their

voices are infringed upon (Baris, 2024; Ruslan, &

Evgenia, 2024).

Although the legal protection systems for the right

to voice in various countries are not yet perfect at the

present stage, some legal measures specifically

targeting AI-generated voices are already being

piloted and promoted. For example, the U.S. state of

Tennessee enacted the Ensuring the Safety of

Portraits, Voices, and Images Act (ELVISAct) in

2024, which expanded the scope of the state's

statutory right of publicity, exposing the behaviors of

artificial intelligence services, internet platforms and

so on, that use artists' voices and portraits to new

liability risks (McCarthy, 2024)

Scholars and legislators in various countries have

produced a wealth of arguments on the legal attributes

of sound. Taking into account the current

development of AIGC technology and the outcome of

China's first AI sound infringement case, the authors

of this article believe that the development of sound

rights and interests has become a booming trend, and

Yang Lixin's view on separate legislation for sound

rights and interests is more realistic and feasible in the

contemporary era.

3.1.2 Does the Voice Have Recognizability?

The recognizability of the voice is a prerequisite for

legal protection. Recognizability can be further

understood as whether the AI work is creative or not,

and whether it can cover the original human voice.

The Beijing Internet Court pointed out on its official

platform that the recognizability of a natural person's

voice means that a specific natural person can be

identified through the characteristics of that voice on

the basis of repeated multiple or long-term listening

by others. Not coincidentally, before the introduction

of AI statute law, the standard of recognizability for

natural person's voice can be analogized to AI-

generated voices. If the voice synthesized by artificial

intelligence enables the general public or the public

in relevant fields to associate it with a specific natural

person based on its timbre, intonation, and

pronunciation style, it can naturally be determined to

be identifiable (Beijing Internet Court Research

Group, 2024). Patel suggests granting copyrights to

AI voice-over models and regarding the outputs as

original works, as a way to fulfill the conditions for

authors of this paper believe that this suggestion is not

applicable to China's national conditions. AI

technology is just at the initial stage in China and

there are relatively few relevant judicial precedents.

Protection of Voice Actors’ Rights and Interests in the Context of AI Speech Synthesis Technology

349

Before the official artificial intelligence act is issued,

AI should not be arbitrarily identified as the creative

subject.

In addition, Wang Shaoxi also proposed that

“when determining identifiability, a distinction

should be made between celebrities and ordinary

people, and the cognitive standards of specific groups

should be taken into account (Wang, 2023)”. In the

author's opinion, the starting point of the argument is

reasonable, but it fails to clarify how to apply it.

Subsequent research is needed to conduct in-depth

interpretations from aspects such as regionality,

social circles, and other directions.

Guaranteeing voice rights and interests through

independent legislation can fundamentally curb the

chaos of infringement and avoid the identification

problems caused by the dilution of voice data. Only

by promptly improving the identification and

protection of the right to voice at the legislative and

judicial levels can the ability to protect the right to

voice in a normalized manner be enhanced.

3.2 Improve the Voice Authorization

System

The AI technology collects the voices of dubbing

practitioners and enters them as training data into the

information database, which will make it more

difficult to define the voices products that are accused

of infringement and more difficult to promote the

process of providing evidence and safeguarding

rights.

Thus, in addition to legislation, people can also

improve the chaos by improving the author-ization

system. Unilateral authorization is a relatively

common authorization method nowadays. For

example, on April 24, the voice actor Qianjing Zhao

announced that he had authorized his voice to the AI

audio series A Record of Mortal's Cultivation to

Immortality produced by TME. Another example is

that after the huge popularity of Ne Zha 2, the voice

actor Yanting Lü used her exceedingly distinctive

voice of Ne Zha to record commercials in ex-change

for commercial benefits. As can be seen from the

above, legal authorization is undoubtedly a good

proof of voiceprint protection.

Conversely, legal authorization can not only bring

commercial benefits but also turn the tide in critical

moments. For example, when the famous voice actor

Guangtao Jiang was unable to participate in voice

recording work due to suspected criminal offenses,

the game project team used the "Anti-Entropy AI"

technology to generate the voice of that character.

This measure can, to the greatest extent, avoid the

subsequent lack of in-game voice re-sources, reduce

the operational risks of the project, and mitigate the

company's losses. However, the one-way

authorization mechanism has always been restricted

by efficiency and is in-sufficient to meet the needs of

economic and social development. Therefore, we

need to draw on advanced domestic and foreign

experiences and create diverse authorization methods

that suit different development models. Concerning

the relatively complete portrait rights authorization

models internationally, it can be specifically divided

into individual authorization or the agency of industry

organizations and the coordination measures of some

authorization agencies. Also, as in the case of the

China Music Copyright Association, the industry

organization agent mode, in various parts of the rights

protection management agencies, is unified

management. Through centralized management,

sound rights holders can save time and energy while

efficiently obtaining commercial remuneration,

allowing industry organizations to drive economic

and social employment and optimize the efficiency of

resource allocation in their business, and ultimately

achieving a “mutually beneficial” authorization

system.

3.3 Construct a Voice Evaluation

Mechanism

3.3.1 What Are the Specific Judgment

Criteria for the Recognizability of a

Voice?

The question of by what criteria the recognizability of

sounds should be judged is a matter that the judiciary

urgently needs to address. The research group of the

Beijing Internet Court believes that it should be

comprehensively considered from two aspects:

subjective criteria and usage methods, supplemented

by whether the general public or the public within a

certain scope can recognize it as the judgment

standard (Research Group of Beijing Internet Court,

2024). Budnik and Evpak proposed a hypothesis for

the legal protection of voice identity, that is, to create

a data identity covering multiple aspects such as voice

parameters and vocal characteristics. On this basis,

trained generative neural networks are used for

identification and comparison, which can provide

more effective legal bases and technical means for

resolving disputes over voice cloning and the

unauthorized use of voices.

ICPLSS 2025 - International Conference on Politics, Law, and Social Science

350

In the first judicial case regarding AI voice

infringement in China, the judge determined through

an on-site inspection in court that the AI voice had a

high degree of consistency with Ms. Yin's timbre,

intonation, pronunciation style and so on. Thus, it was

inferred that the plaintiff's voice rights and interests

extended to the AI voice involved in the case (Beijing

Internet Court, 2023). The above-mentioned method

takes the judge as the subject of appraisal, and the

fluctuating influence on the judgment result due to the

lack of professional knowledge of the lay audience

cannot be excluded. Therefore, the authors of this

article believe that when determining whether a work

that infringes upon others' voices is an act of

infringement, the expert witness system is indeed

necessary. Experts can analyze the similarities in

aspects such as voice characteristics and expression

techniques from a professional perspective, thereby

obtaining more convincing appraisal results.

In addition, in terms of matching the similarity

between the production content and the suspected

source, voiceprint comparison is a common means of

identifying infringement. Voiceprint collection

equipment is introduced to verify the similarity, and

set threshold standards to test the recognizability of

the voice. For example, if the similarity reaches more

than 80%, it may be determined by the judicial

authorities as a substantial similarity, thus

constituting an infringement.

3.3.2 How Should the Responsibilities Be

Determined After the Occurrence of

an Infringement?

Sound products generated through voice processing

are, to a certain extent, similar to musical works.

Therefore, the judgment methods in music

infringement cases can be transferred applications.

Manuelian pointed out that in music copyright

infringement cases, to determine infringement, the

plaintiff needs to prove three elements: having a valid

materials, and the defendant's replication constituting

“inappropriate appropriation" (Manuelian, 1988).

The author of this paper believes that the above-

mentioned judgment ideas can also be applied by

reference in the dubbing industry. Moreover, based

on the practical problem of difficulty in the burden of

proof, it is necessary to refer to the provisions of the

Tort Liability Law of China and reverse the burden of

proof under specific circumstances. That is, the

infringing party shall prove that the data resources

used for training AI do not contain the biometric

information of the infringed party. If the infringing

party is unable to prove itself or has no training

materials, the similarity between the materials

provided by the prosecutor and the generated content

will be used to judge how much to penalize.

Hutiri and Wiebke pointed out that it is necessary

to explore an accountability system for training data.

For example, authenticate the source of training data

and provide creators with an exit mechanism, such as

canceling training data, so as to regulate the

application of voice generation technology from the

source and protect the rights and interests of relevant

personnel (Hutiri, Papakyriakopoulos, and Xiang,

2024). By clarifying the data source and granting

creators control, it is possible to effectively reduce the

illegal collection and use of data and promote the

development of voice generation technology on a

legal and compliant track.

4 CONCLUSION

In the digital era, as an important biological and

commercial resource, the significance of the relevant

norms for the protection of the rights and interests of

voices has become increasingly prominent. This

paper preliminarily explores three measures to protect

the rights and interests of voice through the analysis

of the reality specificity of the dubbing industry and

the theoretical deconstruction of the domestic and

international scope of voice legislation. Protecting

voice rights through legislation is the inevitable

requirement of the artificial intelligence era, and the

possibility provided by the legal and open nature of

personality rights. Improving the sound authorization

mechanism is a due course of action in line with the

development of the times. The construction of a voice

evaluation mechanism is to empower the le-gal

system with scientific and technological means. The

above thoughts provide a little idea for China's

budding sound right provisions and even the AI draft,

but this paper has yet to go deeper into the technical

aspects of voice identification in the sound

assessment mechanism. In the future, it is expected

that the AI technology and the sound protection draft

will continue to develop and improve to better meet

the diversified development of the dubbing industry

and bring more wonderful interpretations to the

audience.

REFERENCES

Baris, A. 2024. AI covers: legal notes on audio mining and

voice cloning. Journal of Intellectual Prop-erty Law &

Practice, 19(7), 571-576.

Protection of Voice Actors’ Rights and Interests in the Context of AI Speech Synthesis Technology

351

Beijing Internet Court Research Group, First-instance

Judgment on the First Case of Infringement of

Personality Rights by AI - AI-generated Voice in

China, published on the WeChat official account

"Beijing Internet Court" ，

https://mp.weixin.qq.com/s/_GxGaG6Q2NYHJWQuO

tMyrQ（uploaded on April 23, 2024）

Civil Judgment, 2023. Jing 0491 Min Chu No. 12142 of the

Beijing Internet Court.

Der Manuelian, M. 1988. The Role of the Expert Witness

in Music Copyright Infringement Cases. Fordham L.

Rev., 57, 127.

Guodong Xu. Draft of the Green Civil Code. Beijing: Social

Sciences Academic Press, 2004: 92.

Hutiri, W., Papakyriakopoulos, O., & Xiang, A. (2024,

June). Not My Voice! A Taxonomy of Ethical and

Safety Harms of Speech Generators. In The 2024 ACM

Conference on Fairness, Accountability, and

Transparency (pp. 359-376).

McCarthy, C. R., 2024. Artificial Intelligence and the

Entertainment Industry. Studies in Law and Busi-ness.

Proposal of a Three-Part Framework. Fordham

Intellectual Property, Media and Entertainment Law

Journal, 34(2), 451.

Research Group of Beijing Internet Court, 2024. Legal

Recognition of the Infringement of Voice Rights and

Interests by AI-generated Voices—Taking the Case of

Yin Mou v. Beijing Certain Intelligent Technology

Company and Others for Infringement of Personality

Rights as an Example. Law Appli-cation, (09): 123-

133, 129.

Ruslan, B., & Evgenia, E. 2024. Human Voice: Legal

Protection Challenges. Legal Issues in the Digital Age,

(4), 28-45.

Schierholz, V., 2019. Götting/Schertz/Seitz,

HandbuchPersönlichkeitsrecht, München.

The Civil Code of Czechoslovakia (in italics), 1981.

translated by Hanzhang Chen, Law Press, 10-11.

Wang, L. M., 2018. On the Legality and Openness of

Personality Rights. Economic and Trade Law Re-view,

(01): 17-27.

Wang, L. M., 2024. On the Legal Protection Model of

Voice Rights and Interests. Journal of Financial and

Economic Law, (01): 3-20, 5.

Wang, S. X., 2023. Interpretation and Application of Voice

Protection in the Era of the Civil Code, Law

Application, 6(44).

Yang, L. X., Yuan, X. S., 2005. On the Independence of the

Right to Voice and Its Protection in Civil Law. Studies

in Law and Business, (04): 103-109.

ICPLSS 2025 - International Conference on Politics, Law, and Social Science

352