ElevenLabs Unveils Preview of Cutting-Edge Music-Generating AI Model

Voice AI startup ElevenLabs is unveiling a new model that transforms prompts into song lyrics, following a strategy similar to Sam Altman's approach during the launch of OpenAI's Sora, its video-generating AI. By engaging users on social media for creative input, ElevenLabs is aiming to generate buzz around its innovative offering.

Founded by ex-Google and Palantir employees, ElevenLabs specializes in machine learning (ML) for voice cloning and synthesis across multiple languages. The company provides a range of tools, including one for dubbing full-length movies, and is now focusing on breaking into the music industry.

Imagine the potential of this AI model: craft a soothing lullaby for bedtime, create a catchy jingle for a marketing campaign, or develop an engaging music intro for your podcast. With the rise of AI music startups like Harmonai, Lyrical Labs, Suno AI, and Loudly, could ElevenLabs be on the brink of producing the next big hit song?

Additionally, users might have the opportunity to sell these AI-generated songs on ElevenLabs’ marketplace, which launched in January. The company's Voice Library allows users to monetize their AI-cloned voices while retaining control over their use and compensation.

However, AI-generated music faces scrutiny. Concerns are raised about the data used to train the model and whether it includes copyrighted material. If copyrighted works were used without permission, this could lead to legal disputes. There's also unease about how such technology may threaten traditional artists' livelihoods, as AI could replicate an artist's unique style and potentially replace human musicians. The fear is that an AI could easily produce a holiday album without the artist's involvement, and the risk of deepfake technology adds another layer of complexity.

As of now, ElevenLabs has not commented further beyond their social media announcements. While the exact maximum length of the songs the AI can produce remains unclear, examples shared by the company's Head of Design, Ammaar Reshi, suggest the capability to generate lyrics for a standard three-minute song.

Updated as of May 10, 2024, at 7:45 a.m. PST: Clarified that ElevenLabs declined to provide further comments.