A cover song made with a cloned voice sits at the intersection of two entirely different legal systems: music copyright law, which governs the underlying song, and an emerging patchwork of voice, likeness, and right-of-publicity protections, which governs the performer’s identity. Getting a mechanical license for the composition does not clear you to clone someone’s voice to perform it. These are separate permissions, and confusing them is the single most common mistake artists make when this topic comes up.
This article is about the legal boundary itself, not a how-to for making these tracks. Globex Music does not distribute AI-generated or AI-voice-cloned music, so treat this as risk-awareness for artists who might be tempted to blend the two, and as clarification for anyone confused about where a legitimate cover ends and a legal problem begins.
What’s the difference between a cover song and a voice-cloned track?
A cover song is a new performance of someone else’s composition — you sing or play it yourself, in your own voice, and a mechanical license (handled automatically by services like Globex Music) pays the songwriter for use of their melody and lyrics. A voice-cloned track uses AI to synthesize a specific person’s actual vocal identity, typically a famous artist, performing words they never recorded. The first replaces the performer; the second impersonates one.
This distinction matters because mechanical licensing law, built for cover songs, has clear statutory mechanisms in the US. Voice cloning has no equivalent statutory license. There is no fixed fee you can pay to legally clone a specific person’s voice — permission has to come directly from that person or their estate, and in most cases it simply isn’t for sale.
Does a mechanical license cover an AI-cloned voice performance?
No. A mechanical license only clears the composition — the underlying melody and lyrics written by the songwriter. It says nothing about whose voice performs it. If you clone Artist X’s voice to sing a cover of a song Artist X never even recorded, you’ve paid for one permission (the composition) while completely skipping another (the vocal identity), which was never for sale through any licensing body in the first place.
This is why voice cloning creates exposure that ordinary covers don’t have. A standard cover, properly licensed, is about as low-risk as music distribution gets. A voice-cloned cover adds a claim vector that no mechanical license, PRO registration, or distributor moderation process can neutralize, because the right being violated isn’t a copyright in a composition — it’s a right in a person.
What legal protections actually cover someone’s voice?
Right of publicity and, in some jurisdictions, specific voice-misappropriation protections. US right-of-publicity law varies significantly by state — California and Tennessee, for example, have particularly strong statutes given how many performers reside or are domiciled there, and Tennessee’s ELVIS Act (2024) explicitly names AI voice cloning as a form of unauthorized use. Federal copyright law, by contrast, doesn’t currently address voice cloning directly at all — this is genuinely unsettled, state-by-state legal territory, not a solved problem with a clean national answer.
The practical implication: a use that might be tolerated in one state could be a clear violation in another, and platforms have been erring toward removal regardless of jurisdiction, because the reputational and legal risk of hosting unauthorized voice clones outweighs the marginal engagement they generate.
Why do platforms remove AI voice-cloned covers so fast?
Because the complaint usually comes from the artist or label directly, not a rights organization working through a standard dispute queue. A mechanical licensing dispute over a normal cover song tends to move through predictable channels — a claim, a review, sometimes a negotiated resolution. A voice clone of a working artist tends to generate immediate public attention, direct label involvement, and platform-level takedowns within hours, because it’s treated as an identity issue, not a licensing technicality.
That speed matters for anyone assessing risk. A rejected cover song submission is a delay you can fix by requesting the right license. A voice-clone takedown is closer to a permanent mark against a catalog, because the underlying use was never fixable through licensing to begin with — there was no license that could have made it legal.
How is this different from an impression, tribute act, or sound-alike singer?
A human singer who sounds similar to another artist, whether through natural vocal similarity or deliberate stylistic mimicry, has performed a long-standing and generally legal practice — tribute acts, impersonators, and sound-alike singers have existed for decades within cover culture. The legal line has historically focused on actual synthesis or reproduction of a specific person’s voice as data, not a human being choosing to sing in a similar style.
This is a meaningful distinction: singing like someone is an artistic choice; cloning their voice is using their identity as an input. Courts and legislators have been far more comfortable with the former than the latter, which is part of why voice cloning specifically has drawn targeted legislation like the ELVIS Act while stylistic impressions haven’t.
What should an artist do instead if they want to cover a song a certain way?
Perform it yourself. Every legitimate cover song still requires only two things to distribute cleanly: your own recorded performance and a mechanical license covering the composition, which Globex Music includes automatically on every release starting at $1, with no separate licensing fee and no annual subscription. That structure — fast moderation, low per-release cost, and royalty payouts from $10 — was built specifically around real cover performances, not synthetic ones.
Compare the cost and risk profile directly. A properly performed cover through Globex Music costs $1 per release with licensing bundled in, versus DistroKid’s $44.99 annual plan, TuneCore’s $24.99 base fee plus per-cover licensing costs and a 20% commission on social platform earnings, or CD Baby’s $9.95 per single plus a 9% royalty commission that continues indefinitely. None of these figures change the calculus around voice cloning, because no distributor’s pricing model includes a path to legally clone another person’s voice — that permission doesn’t exist as a purchasable product anywhere in the current market.
The bottom line
Mechanical licensing solves the composition problem. It has never solved, and was never designed to solve, the identity problem that voice cloning introduces. Artists who want the low cost and speed of cover song distribution — real performance, licensed composition, fast moderation, payouts from $10, no annual fee — already have a clean, well-established legal path. Voice cloning isn’t a shortcut into that same system; it’s a different legal category entirely, one where the permissions required often aren’t available at any price.
Sign up or log in to your dashboard and upload your release now
Latest from the blog
- How a Cover Artist Built a Career Without Writing a Single Original Song
- K-Pop Cover Songs: Licensing Rules Every Artist Should Know
- How to Appeal a Rejected Cover Song Submission
- Cover Song Royalties and Chargebacks: What Artists Should Watch For
- Summer Anthem Covers: How to Time Your Release for Peak Season


