ElevenLabs Music v2.5

• 🎵 Newest Eleven Music model for full text-to-music generation
• 🆕 Builds on Music v2 with improved audio quality and prompt adherence
• 🎯 ElevenLabs reports biggest gains in vocal-led, acoustic genres
• 🔧 Supports Audio Reference, composition plans, and section inpainting
• 📏 Configurable track length, from three seconds up to ten minutes
• 🌐 Vocals across many languages, including English, Spanish, German, Japanese
• 💬 Optional AI-generated lyrics and vocals, or instrumental output
• 🏢 Selectable in the ElevenLabs music interface and the API

ElevenLabs

ElevenLabs is a software company specializing in natural-sounding speech synthesis and audio generation powered by deep learning. The company has established itself as a leading force in AI-driven voice technology, building tools that span text-to-speech,…

ElevenLabs Music v2.5 is the company's newest text-to-music model, which ElevenLabs describes as its best music model yet. It generates complete tracks — instrumentals or full songs with vocals — from natural-language prompts, and is selected by its music v2.5 model identifier in the API or via the model dropdown in the Eleven Music interface. Compared with ElevenLabs' speech-focused systems like ElevenLabs TTS v3 and ElevenLabs Turbo v2.5, or the shorter clips produced by ElevenLabs Sound Effects, Music targets full musical arrangements.

Relative to its direct predecessor ElevenLabs Music v2, the company's documentation says v2.5 "builds on Music v2 with improved audio quality and prompt adherence," while supporting the same feature set — Audio Reference, composition plans, and inpainting. In its announcement, ElevenLabs says tracks sound "more layered and complex" with "more natural-sounding instruments," and that the difference from v2 is widest on vocal-led and acoustic-heavy material such as R&B, soul, hip hop, rock, metal, and orchestral or cinematic music. Music v2 itself remains available.

The underlying capabilities carry over from the v2 generation, which ElevenLabs positioned as an upgrade over the original ElevenLabs Music in prompt adherence, composition, multilingual output, and vocal delivery. Practical controls include audio references of roughly 30 seconds to steer mood, instrumentation, and production style; section-by-section composition plans with their own lyrics, styles, and durations; regeneration of individual sections; and durations configurable from three seconds to ten minutes.

ElevenLabs states the model is built with artists, labels, and publishers and cleared for commercial use under its music terms, with uploaded reference audio screened for copyright compliance. Note that no independent third-party benchmark results are published for this model, so quality claims here are the provider's own.