The voiceover engine behind most faceless YouTube and audiobook stacks. Voice cloning, multilingual, low-latency streaming.
POST /v1/speech-to-speech/{voice_id} paths are deprecated in favor of POST /v1/voice-changer/{voice_id}. Legacy paths continue to work but should be migrated. SDKs from v2.75.0 onward call the new paths automatically.
ElevenAgents can now run supported LLMs on a fast service tier for higher token throughput and lower latency, trading a per-token cost premium. Set service_tier=fast in prompt configuration; availability varies by model.
Text to Speech and Text to Dialogue endpoints now accept clone_strength parameter (0–1) to control how closely v4-finetuned clones follow training samples. Recommended value is 0 for most use cases.
Agent knowledge base website crawls can now exclude up to 50 URL patterns to skip irrelevant content. Improves knowledge base quality for deployed agents using web crawl sources.
Text-to-Speech and Text-to-Dialogue APIs now accept clone_strength (0 to 1) to control how closely Professional Voice Clones fine-tuned for v4 models follow their training samples, enabling developers to fine-tune voice synthesis behavior.
ElevenAgents now supports a fast service tier for supported LLMs that trades higher per-token costs for improved token throughput and lower latency. Developers can configure service_tier in agent prompt configuration.
ElevenAgents knowledge base crawls can now specify up to 50 URL patterns to exclude during site indexing via exclude_patterns parameter, helping developers prevent sensitive or irrelevant content from entering agent knowledge bases.
Workspace service account API keys can be configured with expiration times between 15 minutes and 90 days in the future, improving security and credential lifecycle management for automated integrations.
Voice Changer moved from /v1/speech-to-speech/{voice_id} to /v1/voice-changer/{voice_id}. The old paths remain functional but are deprecated. Python and JavaScript SDKs updated from v2.75.0; method names unchanged for now.
Knowledge base website crawls can now skip URLs matching up to 50 exclude patterns. Configure exclusions via Create crawl job with exclude_patterns parameter or in the dashboard's Exclude patterns field.