ElevenLabs Releases SFX v3 Sound Effects Model with Improved Multi-Sound Prompt Handling
ElevenLabs launches SFX v3, enhancing reliability for generating multiple sounds from a single prompt and improving music/vocal quality.
AI audio company ElevenLabs released its latest sound effects model, SFX v3, on October 29, 2026. The model is now available in the ElevenCreative platform and via ElevenAPI.
According to the official ElevenLabs blog, the primary improvement in SFX v3 is its ability to more reliably handle single prompts that describe multiple sounds. For example, users can request crowd cheering, a ball dropping, and subsequent booing in one instruction, and the model better integrates these elements. Additionally, the new version generates more convincing music and vocals, and reduces unwanted artifacts in sounds like gongs, chimes, and whistles, resulting in more natural overall audio quality.
ElevenLabs states that these improvements aim to help video and game creators explore scene sounds more efficiently without trial and error. Developers can integrate this feature into their own products through ElevenAPI.