Eleven Music v2.5

Eleven Music v2.5 is a flagship text-to-music generation model developed by ElevenLabs. Designed to create studio-grade full songs and instrumentals from natural language prompts, it builds upon its predecessor with enhanced audio quality, richer melodies, and improved prompt adherence. The model excels at producing more layered arrangements and instruments that sound closer to live takes, particularly across vocal-led and acoustic-heavy genres like R&B, soul, hip-hop, rock, and cinematic orchestration.

Capabilities and Workflow

The model supports advanced workflows including Audio Reference generation, where users can upload a short track to guide the style, tempo, and mood without copying the audio directly. It also features inpainting and section-by-section editing capabilities, allowing creators to refine individual parts, change lyrics, or alter bridges and choruses without disrupting the rest of the continuous track. Multilingual support enables native-sounding vocals and lyrics across multiple languages including English, Spanish, German, and Japanese.

Prompting Guidance

To achieve optimal results with Eleven Music v2.5, creators should use descriptive natural language inputs detailing genre, mood, instrumentation, and structural elements such as tempo and genre transitions. Utilizing custom or curated Finetunes allows users to steer the stylistic identity of the output. When building complex compositions, starting with a shorter duration and iteratively adding sections via the editor helps maintain musical coherence throughout the track.