AI voice-swapping is coming: what Germany's GEMA rulings mean for working musicians
Spotify reportedly plans AI remixes and voice-swapping for 2027. Here is how Germany's court rulings and opt-in deals affect performers.
In this article
Two pieces of music-business news from the past few days point in the same direction. According to a Financial Times report covered by Complete Music Update on 30 October, Spotify is aiming to launch a paid add-on in early 2027 that would let listeners remix and cover existing tracks with AI, including a "voice swapping" option that imitates a specific artist's vocals. And on 2 November the same outlet looked at an AI-generated blues act reaching number two on the UK singles sales chart, reigniting the debate about where AI music belongs. For working musicians across Europe, the practical question is no longer whether AI will touch your music, but on what terms.
Germany has already drawn some lines
The clearest legal signals so far come from Munich. On 31 July 2026 the Munich Regional Court largely sided with GEMA, the German collecting society, against the AI music service Suno. The court found that training on GEMA repertoire and generating outputs that closely matched protected songs infringed copyright, held the AI provider (not its users) responsible, and said licence fees and damages are owed. The judgment is not final and Suno has said it is weighing an appeal. An earlier GEMA case against OpenAI over song lyrics is also still before the appeal court.
At the same time GEMA has started selling a licensed training dataset to AI developers. In other words, the message from Germany is "license it, don't take it", and collecting societies elsewhere in Europe are watching closely.
Why opt-in matters for performers, not just stars
The reported Spotify product depends on artists and songwriters agreeing to take part, especially where their voice is involved. Labels are said to be approaching artists one by one, and revenue-sharing proposals reportedly range from streaming-style royalty rates to a 50/50 split. None of this is final, and Spotify has a history of announced premium tiers arriving late or not at all, so treat the details with caution.
Still, the principle is useful for everyone: your voice, name and recordings are rights you can grant or withhold. That applies whether you are signed to a major or you are a wedding singer whose demo videos sit on social media. Cover bands and session players should also note that their own performances of other people's songs are recordings with value, and could be scraped or cloned like anything else.
The chart story is a warning about fraud, not quality
The AI act's chart placement was modest once you look closely: it appeared on a sales-only chart and not in the main chart that includes streaming. The bigger worry raised in the coverage is that AI tools can make streaming manipulation and impersonation easier, and inflated streams take money out of the shared royalty pool that pays real musicians. Spotify says it focuses on spam and impersonation rather than on how music is made, and offers a beta tool that lets artists approve releases before they appear on their profile.
What to do
Read your contracts for AI clauses. Before signing a label, distribution or session agreement, look for wording about AI training, voice models or "new technologies", and ask for opt-in rather than blanket consent.
Register your works properly with your collecting society (GEMA, SACEM, SIAE, SABAM, AEPI and others) so you are visible when licensing money is distributed.
Protect your artist profile. Check streaming profiles regularly for tracks you did not release and use the platforms' reporting and approval tools.
Keep your originals and session files organised and dated; they help prove authorship if a dispute arises.
Talk about it with clients. If an event host wants to film your set, agree in writing how the footage can be used.
Get advice before signing anything unusual — a musicians' union or a music lawyer can check AI terms quickly.