Written from 15 named sources · Oct 16 · first result A research companion to VO Infusion Episode 13, "Vocal Intellectual Property." Researched with Pig Knuckle and reviewed by Deb Munro and Keith Norton, with sources for every claim. Federal copyright law protects a specific script or a fixed sound recording, but it does not treat your raw vocal cords or vocal timbre as registrable property [10, 11]. State right of publicity statutes and common law protect personal identity against unauthorized commercial impersonation, while contract law serves as your primary operational defense against unauthorized model training [10, 11]. Tennessee's ELVIS Act explicitly codified voice as a protected property right in July 2024, California has enacted specific rules governing digital replica agreements, and the federal NO FAKES Act remains a proposed bill rather than enacted law as of October 2026 [7, 12, 13]. Canadian voice actors rely on provincial privacy statutes, common law appropriation of personality, and ACTRA collective bargaining agreements rather than an overarching federal right of publicity. Biometric litigation under Illinois BIPA treats a voiceprint as an algorithmic mathematical feature vector capable of identifying a person, distinguishing technical extraction from simple audio playback [2, 4, 6]. An "in perpetuity" clause removes an expiration date, but the broader legal danger stems from surrounding grants covering synthetic digital replicas, model training, and all media hereafter devised. Legitimate synthetic voice licensing pathways already exist outside unvetted platforms, exemplified by SAG-AFTRA's August 2024 agreement with audio advertising marketplace Narrativ [9, 14, 15]. Under US law, can a voice be owned? A large company once asked me what a clone of my voice would sound like. We fed audio into their system, spent about an hour tweaking parameters to get the clone somewhere acceptable, and when I stepped up to the microphone myself, my natural read of that same corporate legal copy was quicker and had far more depth. That session stuck with me because it exposed the gap between capturing a performance and mimicking a human being. It also forced an uncomfortable legal question: do we actually own the sound that comes out of our throats? Under United States law, the short answer is no... at least not in the way you own a car, a registered trademark, or a copyrighted musical composition. Your vocal cords are human anatomy, not patentable software. Legal protection for voice actors is assembled across four distinct legal frameworks that perform very different tasks. Copyright law requires an original work of authorship fixed in a tangible medium of expression. That means copyright protects the recorded WAV file you deliver to a client or the script crafted by a copywriter. It does not protect your underlying vocal characteristics. If an advertiser hires someone who sounds like you, or feeds existing recordings into an algorithmic synthesizer to generate new dialogue from scratch, you cannot sue for copyright infringement of your vocal tone alone [10, 11]. Under Section 114(b) of the Copyright Act, exclusive rights in a sound recording do not extend to the making of another recording that consists of an independent fixation of sounds, even if those sounds imitate the copyrighted recording [10]. The practical limits of copyright surfaced in Lehrman v. Lovo, Inc., a putative class action brought in the Southern District of New York by voice actors Paul Lehrman and Linnea Sage [1, 10]. They alleged that an AI software company hired them via Fiverr under the representation of "academic research" scripts, only to feed those takes into its speech generator "Genny" to market commercial clones dubbed "Kyle Snow" and "Sally Coleman" [10, 11]. In July 2025, Judge J. Paul Oetken dismissed their primary copyright output claims and contributory infringement theories, reaffirming that a voice itself falls outside the subject matter of federal copyright [10, 11]. While Sage's claim over the unauthorized reproduction of her actual audio in promotional investor presentations survived, the court made clear that new synthetic dialogue imitating a voice does not constitute copyright infringement of the original sound recording [10, 11]. Following Lovo's subsequent Chapter 11 bankruptcy filing, the action was automatically stayed in May 2026, yet the judicial opinion remains a baseline ruling on vocal ownership [8, 10]. That identity layer belongs to the right of publicity. The right of publicity is predominantly a state-level doctrine that stops commercial entities from exploiting your name, image, likeness, or persona without consent. Long before generative models, two landmark Ninth Circuit rulings laid this foundation for our industry. In Midler v. Ford Motor Co. (1988), Ford hired Bette Midler's former backup singer to mimic Midler's performance of "Do You Want to Dance" for a commercial after Midler declined the job [7]. The court held that when a distinctive voice of a professional singer is deliberately imitated to sell products, the seller appropriates identity [7]. Tom Waits expanded that protection in Waits v. Frito-Lay, Inc. (1992), securing a substantial jury verdict after an agency hired a vocalist to imitate his gravelly rasp for a Doritos radio commercial [7]. Publicity rights help voice actors, but they carry notable limitations. Historically, they required you to prove that your voice is commercially famous and immediately recognizable to the public. That standard leaves emerging talent and non-celebrity voice actors in a vulnerable legal position. Trademark law, governed nationally by the Lanham Act, rarely fits regular commercial bookings. A trademark indicates the commercial source of goods or services. Unless your voice functions primarily as a brand mark that tells consumers who manufactured a product, trademark claims will fall flat. In Lehrman v. Lovo, the court dismissed both the false association and false advertising claims under Section 43(a) of the Lanham Act [10, 11]. Judge Oetken noted that the actors' voices were used as the product itself rather than as source identifiers, warning that using trademark law to police voice imitations would distort the boundaries established by copyright and publicity law [10, 11]. Contract law remains your most practical, enforceable line of defense. While the court in Lehrman narrowed the statutory intellectual property theories, it firmly sustained the performers' breach of contract claims [10, 11]. The online chat exchanges between the actors and Lovo's agents, combined with the platform terms, established material terms limiting the recordings to research use [10, 11]. If an agreement explicitly licenses audio solely for an internal corporate video and bars model training, the client cannot legally feed those files into a neural network without breaching the agreement [10, 11]. Because the legal system does not give you automatic property ownership over the sound of your voice, you must define those property boundaries inside every contract you sign. What do the ELVIS Act, California digital replica laws, and the NO FAKES Act do? State legislatures have recognized that common-law precedents from the radio and television eras cannot match the speed of generative speech synthesis. Professional voice acting did not start two or three decades ago; it traces directly back to 1920s radio dramas and 1930s animation studios. For a century, a performer delivered a performance in a studio and received compensation for that fixed session. Modern cloning severed that relationship, triggering a wave of statutory updates. Statute / Proposal Jurisdiction Status as of October 2026 Primary Protection Provided Key Limitation ELVIS Act Tennessee Enacted (Effective July 1, 2024) [7, 12] Adds "voice" as an explicit property right; penalizes clone creators and hosting tools [7, 12]. State-level statute; requires a jurisdictional connection to Tennessee [7]. AB 2602 & AB 1836 California Enacted (Signed September 2024) [7, 13] AB 2602 voids vague digital replica contract clauses; AB 1836 protects deceased performers [7, 13]. Living performers can still knowingly sign replica rights away with counsel [7]. NO FAKES Act US Federal Proposed Bill (Reported to Senate calendar June 2026) [7, 12, 13] Creates a nationwide property right against unauthorized digital voice replicas [7, 13]. Not enacted law; cannot be cited as an active cause of action in negotiations or court [7, 12]. Tennessee moved first by passing the Ensuring Likeness Voice and Image Security (ELVIS) Act, which took effect on July 1, 2024 [7, 12]. The law amended Tennessee's Personal Rights Protection Act to make a person's voice an explicit property right alongside name, photograph, and likeness [7, 12]. The ELVIS Act eliminated the traditional barrier requiring an individual to prove commercial celebrity before seeking statutory relief [7, 13]. It also created direct civil liability for creating or distributing unauthorized vocal simulations, while targeting platforms and software services whose primary purpose or effect is producing identifiable voice clones [7]. California responded in September 2024 with Assembly Bills 2602 and 1836 [7, 13]. AB 2602 targets personal service contracts, rendering provisions unenforceable if they permit the creation and use of a digital replica in lieu of the performer, unless the agreement includes a specific description of intended uses and the performer was represented by legal counsel or a union representative [7, 13]. AB 1836 updated protections for deceased performers, establishing statutory damages starting at $10,000 for exploiting a digital voice or likeness replica without estate consent [7]. Federal legislation remains unfinished on Capitol Hill. The Nurture Originals, Foster Art, and Keep Entertainment Safe (NO FAKES) Act was introduced to establish a uniform, nationwide property right protecting individuals against unauthorized digital voice and likeness replicas [7, 12, 13]. A revised bipartisan version was reported out of the Senate Judiciary Committee and placed on the Senate Legislative Calendar on June 24, 2026 [7, 12, 13]. As of October 2026, the NO FAKES Act has not been voted on by the full Senate or passed by the House of Representatives [7, 12]. A bill on a legislative calendar provides zero immediate legal protection when you review an invoice or catch an unauthorized clone [7]. Until federal law creates a uniform standard, voice talent must evaluate their rights under a fragmented legal map. If you live in Illinois, you evaluate biometrics; in California, you review digital replica contract rules; in Tennessee, you rely on the ELVIS Act [7, 13]. For talent working everywhere else, your rights are determined almost entirely by the specific terms in your agreements. How does Canada treat personality rights, privacy, and voice protection? North of the border, the legal picture shifts. Canada lacks a single, overarching federal right of publicity comparable to state statutes in Tennessee or California. Protection for voice actors is divided between provincial privacy legislation, common-law torts, and collective bargaining agreements. In British Columbia, Manitoba, Newfoundland and Labrador, and Saskatchewan, protection is codified within provincial Privacy Acts. These statutes make it an actionable statutory tort to use an individual's name, portrait, or voice for advertising or promotional purposes without consent. In Ontario and other common-law provinces, courts recognize the tort of "appropriation of personality." Established in foundational cases like Athans v. Canadian Adventure Camps Ltd. (1977), the tort establishes that an individual holds an exclusive proprietary right in the commercial marketing of their personality, image, and voice. Quebec handles personality exploitation through civil law protections grounded in the Civil Code of Québec and the Quebec Charter of Human Rights and Freedoms. For Canadian voice actors, the most practical everyday protection comes through union collective bargaining rather than civil courts. The Alliance of Canadian Cinema, Television and Radio Artists (ACTRA) has established contractual guardrails covering generative artificial intelligence and digital replicas across its agreements. Under these frameworks, an engager cannot build a synthetic replica or feed a member's recordings into machine learning systems without separate, informed consent and negotiated compensation. If you are a Canadian performer booking non-union commercial or corporate narration, American right of publicity statutes will not automatically shield you. If your contract designates Ontario or British Columbia as the governing jurisdiction, you cannot cite the ELVIS Act to resolve an issue [7]. You must rely on provincial privacy rules, common-law personality claims, and explicit restrictive covenants in your paperwork. What does "in perpetuity" actually mean in a voiceover agreement? Deb sees "in perpetuity" attached to nearly every pay-to-play audition script that crosses her desk, and every working performer knows that sinking feeling. In contract law, "in perpetuity" means forever. It contains no expiration date, no renewal window, and no sunset clause. In traditional broadcasting, an in-perpetuity buyout applied to a single, static deliverable: a corporate client paid an upfront fee to run one specific training video or explainer indefinitely without paying residuals. In an automated production pipeline, buyers now attach that phrase to broad media rights, transforming a routine broadcast buyout into an enduring license to simulate your voice forever. Examine how these boilerplate phrases interact inside standard contracts: "The right to use, modify, adapt, reproduce, and create derivative works from the performance..." "...in all media, now known or hereafter devised, worldwide, in perpetuity." "...for internal testing, machine learning evaluation, algorithmic development, or AI training." "...including the right to create, license, or exploit a synthetic digital replica of the artist's voice." "...with full rights of assignment, transfer, and sublicensing to corporate affiliates and third parties." When those clauses appear together, the client is not purchasing a voiceover read; they are acquiring the legal right to build your digital replacement. If you sign that paperwork, you grant permission to train an AI model on your audio files, generate scripts you never saw, and distribute that clone across future media formats without paying you another dollar. Performers do not have to accept these terms as standard operating procedure. You can strike the offending sentences or attach a standard protective rider. The National Association of Voice Actors (NAVA) created an open-source AI rider that working voice talent use to explicitly prohibit unauthorized synthetic media generation and AI training. My agent reviews and negotiates my major contracts to ensure generative rights are removed or priced at proper commercial value, while Deb uses language models to scan dense audition agreements for buried machine learning clauses. A language model serves as a practical magnifying glass for highlighting dangerous terms on a phone screen, but software cannot negotiate on your behalf. When a lucrative commercial contract or a confusing digital replica clause lands on your desk, have a talent agent, an entertainment lawyer, or your union inspect the wording before delivering your files. It is equally necessary to separate unauthorized voice theft from legitimate synthetic licensing. ElevenLabs is not the only platform connecting voice talent with commercial buyers. In August 2024, SAG-AFTRA signed a notable agreement with AI audio platform Narrativ [9, 14, 15]. The agreement allows union voice performers to license synthetic digital replicas for digital audio advertisements while retaining approval over every individual campaign, establishing their own pricing floors above union minimums, and securing health and retirement contributions [9, 14, 15]. Other platforms, including Replica Studios, Respeecher, and Veritone, also license voices under defined contractual terms [9]. Synthetic voice licensing is developing into an organized commercial sector, and the difference between exploitation and a viable business deal depends entirely on informed consent, clear boundaries, and fair compensation [9, 14, 15]. What does a voiceprint mean legally under biometric privacy laws like Illinois BIPA? During Episode 13, I mentioned that protecting our vocal cords might ultimately hinge on our vocal prints. That intuition is grounded in an active front in privacy litigation: state biometric privacy legislation. Under the Illinois Biometric Information Privacy Act (BIPA), private companies cannot collect, capture, purchase, or obtain a person's biometric identifier without giving written notice, stating the specific purpose and duration of collection, and securing a signed written release [2, 5]. BIPA explicitly lists a "voiceprint" as a protected biometric identifier alongside retina scans and fingerprints [2, 5, 6]. A voiceprint is not an ordinary MP3 or WAV recording [2, 6]. In legal and technical proceedings, a voiceprint is a mathematical, algorithmic mapping of your unique physiological vocal tract, converted into numeric feature vectors capable of identifying a speaker [2, 6]. Security systems use voiceprints for user authentication, and machine learning architectures extract acoustic feature representations to analyze speech or synthesize new dialogue [2, 3, 6]. This distinction has driven major courtroom battles. On January 29, 2026, an Illinois court certified a class action against Apple on behalf of Illinois residents whose voiceprints or biometric feature vectors were computed and collected through Siri without statutory BIPA disclosures [2, 4]. Class litigation has challenged other technology platforms as well [3]. In June 2026, a federal judge unsealed an order in Delgado v. Meta Platforms, Inc. denying Meta summary judgment, ruling that a genuine factual dispute existed over whether Meta's internal pipelines extract identifiable biometric voice data from Facebook and Messenger audio [6]. BIPA provides statutory damages of $1,000 for each negligent violation and $5,000 for each intentional or reckless violation, alongside attorney fees [2, 4, 5]. While an August 2024 amendment (Public Act 103-769) clarified that repeated collections of the same identifier using the same method count as a single violation, the financial exposure across large user bases remains substantial [5]. However, BIPA is not an automatic remedy for voice actors. BIPA is an Illinois state statute [5]. Unless the processing occurred within Illinois or involved Illinois residents, talent located in other states or Canadian provinces cannot easily invoke it [5]. Furthermore, biometric privacy actions challenge the extraction of the identifying mathematical vector, not the artistic exploitation of a performance [2, 6]. Biometric privacy safeguards physiological data, publicity law protects personal identity, and contract law safeguards professional labor. Understanding which tool addresses your specific situation is critical. What this means for your voice business Audit your contract terms and platform agreements this month. Review every casting platform profile, agency agreement, and client estimate form you use. Search directly for terms like "derivative works," "machine learning," "synthetic replica," "hereafter devised," and "in perpetuity." If an agreement includes model training rights without separate compensation, cross them out or decline the booking. Add an explicit AI non-consent rider to every estimate and invoice. Insert clear terms stating: "Client is granted a license for finished audio playback only. Client is explicitly prohibited from using delivered audio files to train, fine-tune, or develop artificial intelligence models, speech synthesis systems, or synthetic digital replicas without a separate written agreement." Separate your rates for recording labor versus synthetic replica licensing. If a client asks to build an internal corporate clone, do not bundle that into a standard session fee. Divide your pricing into three distinct categories: studio labor for recording training data, a term-limited synthetic license fee, and usage fees for generated output. If an AI platform approaches you, study structures like the SAG-AFTRA Narrativ agreement for standards on consent, compensation, and campaign approvals [9, 14, 15]. Preserve written paper trails for every booking. Create a dedicated digital folder for every session containing initial casting posts, email conversations, scripts, signed agreements, and payment records. In Lehrman v. Lovo, the actors' breach of contract claims survived a motion to dismiss because their written Fiverr messages proved the audio was provided solely for research purposes [10, 11]. Contemporaneous records serve as your primary evidence in any legal dispute. Test tools cautiously and monitor new production formats. If you experiment with speech synthesis tools or consider Deb's idea of displaying a clone sample next to your real read on your website to demonstrate human depth, review the software terms of service first. Ensure that uploading your audio does not grant the developer a perpetual license to use your voice for model training. Monitor emerging commercial markets as well: vertical microdrama apps backed by investor funding produce series with production budgets typically ranging from $15,000 to $300,000 (reaching up to $600,000 for premium productions), while developments like Paramount's 1888 Studios project in Bayonne, New Jersey, highlight continued investment in production infrastructure. Expanding media channels create real voiceover work if you protect your rights from the start. Escalate complex digital replica contracts to professional representatives. When an audition agreement includes ambiguous synthetic rights or broad media buyouts, do not guess. Consult your talent agent, contact your union representatives at SAG-AFTRA or ACTRA, or hire an entertainment attorney experienced in digital media. Retaining qualified counsel for thirty minutes of contract review is far less costly than accidentally licensing away the rights to your voice for the rest of your career. Go further U.S. District Court Opinion in Lehrman v. Lovo, Inc. (2025): The primary judicial ruling examining voice cloning across copyright, right of publicity, and contract law [10]. SAG-AFTRA Artificial Intelligence Resources and Narrativ Pact: An established framework showing how unions structure performer approvals, wage minimums, and health contributions for synthetic audio [9, 12, 14, 15]. National Association of Voice Actors (NAVA) AI Resources: Practical open-source contract riders designed by working voice actors to remove unauthorized AI training terms. Illinois General Assembly - Biometric Information Privacy Act (740 ILCS 14/): The statutory text governing voiceprints and mandatory notice-and-consent protocols for biometric collection [2, 5]. ACTRA National Commercial and Independent Agreements: Essential guidance for Canadian voice performers outlining collective bargaining protections against synthetic replacement. About VO Infusion VO Infusion is a podcast about artificial intelligence and the voice acting industry, co-hosted by voice actor and coach Deb Munro and voice actor Keith Norton. Episodes cover voice cloning and licensing, legal protections for performers, prompt engineering for voice actors, AI tools for the business side of voiceover and why human performance still matters. Listen wherever you get podcasts or on YouTube at @VOInfusion, and find more at