I recently stumbled upon an app that claims to create AI-generated voices of rappers and artists such as Drake and The Weeknd. Not only that, but it also writes original lyrics and composes entire songs using artificial intelligence. Curious and somewhat skeptical, I decided to put it to the test.
What the App Promises
The app markets itself as a tool for music enthusiasts and creators alike. Users can select a voice model trained on hours of an artist's vocal recordings, input a topic or mood, and the AI generates both lyrics and a vocal performance that mimics the chosen artist's style, cadence, and tone. The app also offers instrumental backing tracks produced by AI, meaning you can theoretically generate a complete song in minutes.
The Test: Recreating a Drake Vibe
I started with a Drake voice model. I provided a simple prompt: "Write a song about a late-night drive through Toronto, with romantic overtones." Within seconds, the app produced a two-verse track. The lyrics were indeed about cruising through a rainy city, referencing landmarks like the CN Tower and Yonge Street. The vocal delivery, while not perfect, carried Drake's signature melodic flow and slight breathiness. It was eerily close.
Next, I tried The Weeknd. I asked for a song with a dark, synth-heavy vibe about lost love. The AI output had the high-pitched, ethereal quality associated with the Weeknd, complete with ad-libs and vocal runs. The instrumental featured a moody synth pad and a driving beat reminiscent of his "Blinding Lights" era.
How Does the Technology Work?
These AI voice-cloning tools rely on deep learning models trained on massive datasets of an artist's vocals. Using a technique called text-to-speech synthesis with a conditioning mechanism, the AI learns to map textual lyrics to the unique timbre, pitch, and rhythm of a specific voice. Advances in generative adversarial networks (GANs) and diffusion models have made these outputs increasingly realistic. Some apps also employ a separate model for melody generation and harmonization.
Ethical and Legal Concerns
The ability to clone an artist's voice without permission has sparked heated debates. In 2023, a viral song allegedly featuring AI-generated vocals of Drake and The Weeknd on the track "Heart on My Sleeve" led to record labels quickly issuing takedowns. Universal Music Group (UMG) stated that the song violated copyright and infringed on the artists' rights. Since then, platforms like Spotify and Apple Music have updated their policies to remove AI-generated content that mimics real artists without authorization.
However, the app I tested operates in a gray area. It claims to provide only "inspiration" and does not allow downloading or publishing of full songs without a license. Yet, the boundaries remain fuzzy. As of 2026, the U.S. Copyright Office has not issued clear guidelines on AI-generated voice clones. Some artists, like Grimes, have openly embraced similar technology, allowing fans to use her voice in exchange for revenue sharing. But others, such as Drake, have threatened legal action.
Broader Implications for Music Creation
Beyond ethics, AI song generators democratize music production. Anyone with a smartphone can create a demo that sounds professional, potentially lowering barriers for aspiring artists. Yet, it also threatens the livelihoods of session singers, producers, and even established artists who rely on their unique voices as trademarks.
During my test, I also experimented with the app's lyric-writing capabilities. The AI not only wrote coherent verses but also structured them with hooks, bridges, and ad-libs. For example, when I asked for a country song about a cowboy in space, it produced clever lines like "My horse is a rocket, my hat's made of steel, across the Milky Way I ride for a real deal." The creativity surprised me, though it sometimes used clichés.
Quality Check: Is It Convincing?
The Drake and Weeknd voice models were convincing enough to fool casual listeners. I played the AI-generated Drake track for three friends without revealing its origin. Two guessed it was an unreleased leak; one thought it was a deepfake but wasn't sure. After telling them, they were shocked by the realism. However, the audio lacked the subtle imperfections and emotional nuance of a human performance. The pronunciation of certain words was slightly off, and the dynamic range felt flat at times.
For the Weeknd model, the high notes were impressive, but the app struggled with live-sounding breath control. Still, considering this is a consumer-grade app with no studio-level processing, the results were remarkable.
The Future of AI in Music
Industry experts predict that within the next few years, AI-generated music will become indistinguishable from human-created content. Laws are likely to evolve, possibly requiring clear labeling of AI-generated songs and licensing agreements for voice clones. Some artists are already selling official voice packs to fans for non-commercial use. The test verified that the technology is mature enough to produce passable songs, but it also highlighted the need for responsible deployment.
As I wrapped up my experiment, I reflected on the paradox: the same technology that could enable a home producer to sound like a superstar also challenges the very concept of artistic identity. The app worked—it generated credible Drake and Weeknd songs—but whether that's a boon or a threat depends on how we choose to regulate and embrace it.
In the coming months, more artists are likely to confront this reality. Labels are investing in their own AI tools to protect their catalogs while exploring new revenue streams. Meanwhile, startups like the one behind this app continue to push boundaries, often staying one step ahead of the law. For now, my test confirms that AI song generators are not a gimmick; they are a disruptive force that will reshape the music landscape for years to come.
Source: MSN News