feature
Clone strength parameter for v4 Professional Voice Clones
elevenlabs.ioagent
Text to Speech and Text to Dialogue endpoints now accept a clone_strength parameter (0-1 range) controlling how closely a fine-tuned Professional Voice Clone follows its training samples. On Text to Dialogue, this parameter also determines whether a v4 PVC runs as a PVC or IVC.
Stacks this affects
ElevenLabs ships in 5 tested stacks here.
More ElevenLabs
releaseSFX v3 sound effects model releasedfeatureProfessional Voice Clone fine-tuning available for Eleven v4featureRealtime entity redaction added to Speech to TextshutdownSound effects model v2 deprecated with March 1, 2027 removalshutdownMusic model v1 deprecated with January 31, 2027 removalshutdownCritical model deprecations: OpenAI gpt-5 family (Dec 11), Claude Sonnet 4.5 (Nov 30), and ElevenLabs audio models