ElevenLabs has launched Music v2.5, an upgraded audio generation model available through both its web application and developer API. According to the company, the new version produces fuller, more natural audio compositions. In a blind preference test comprising 47,885 comparison pairs, listeners favored v2.5 over prior iterations, showing strong preferences in R&B, Soul, Hip-Hop, Rock, and orchestral genres. The legacy v2 model will remain accessible alongside the new release.
The service offers tiered access options: free users receive five lossless downloads per day with mandatory attribution, while Pro subscribers get 400 monthly downloads. Users retain full rights to generated tracks, and commercial usage is permitted depending on industry and context. ElevenLabs implemented strict content controls that block downloads of tracks based on existing artists’ songs and prevent direct voice or style imitation.
Regarding data sourcing, ElevenLabs stated that Music v2.5 was trained on licensed music and stems, differentiating its approach from competitor Suno, which faces ongoing copyright litigation. The company noted that its recent licensing agreement with Universal Music Group applies strictly to future, separate products rather than the Music v2.5 release.
Why it matters
API availability allows developers to integrate licensed commercial music generation directly into applications.
Training on licensed stems offers enterprise users lower legal risk compared to competitors facing copyright lawsuits.
Source: the-decoder.com
