Daily AI news, sourcedOfficial news, first rumor reports and substantive progress. Each entry keeps the information published at that time.
- ElevenLabs details Cue, a forced-alignment model for character-level audio timing
ElevenLabs research introduces Cue, a forced-alignment model it says places every character on the audio timeline with more precision than other aligners it tested, so captions and words land at the right time.
- ElevenLabs documents a Realtime Translation WebSocket API for enterprise workspaces
ElevenLabs' docs describe a Realtime Translation WebSocket at /v1/translation/realtime that streams translated speech from source audio into a target language. The API is currently for enterprise workspaces only and supports configurable TTS models, eagerness and input/output audio formats.
- ElevenLabs publishes an OpenAI Realtime API compatibility guide for Eleven Agents
ElevenLabs published a developer guide on connecting Eleven Agents with OpenAI's Realtime API, covering how to wire realtime voice sessions through its agents stack.
- ElevenLabs opens a limited alpha of Eleven Realtime Translation for live voice translation in calls
ElevenLabs, which launched Eleven Realtime for voice agents this week, is opening a limited alpha of Eleven Realtime Translation, which adds live voice translation to calls between people.
- ElevenLabs launches Instant Soundtracks, scoring a video with original music and timed sound effects in one pass
ElevenLabs' Instant Soundtracks takes an uploaded video, generates an original score and places sound effects at matching moments, optionally matching the character of the original audio. It is available in ElevenCreative, Eleven Music and the ElevenAPI.
- ElevenLabs says its own ElevenAgents support agent resolves over 22,000 tickets a month and 65% of tickets end to end
ElevenLabs runs its customer support on an ElevenAgents agent that resolves more than 22,000 tickets a month without involving a person, resolved 65% of tickets end to end last month and cut average first response time from hours to 60 seconds; 70% of the conversations it handled were rated positively. The team turned its existing procedures for refunds, billing changes, voice clone questions and other categories into agent Procedures, using Structured Procedures for critical steps like refunds, and added webhook tools to look up account data. It tested about 50 common scenarios with simulation tests and ran the agent in shadow mode on production traffic before deploying it.
- Baroness Campbell gives a House of Lords speech using an ElevenLabs AI recreation of her voice
ElevenLabs says Baroness Campbell of Surbiton, who has spinal muscular atrophy, delivered a speech in the UK House of Lords using an AI recreation of her voice built by ElevenLabs, which it calls a first for the UK Parliament. The voice replicates the tone, rhythm and pitch of her original voice. ElevenLabs says its technology has restored the voices of more than 11,000 people and that it has committed to help one million people living with voice loss.
- ElevenLabs lets ElevenAgents teams merge and search triage tickets, and adds success or failure outcomes to workflow End nodes
ElevenLabs' October 12 changelog adds a Merge tickets endpoint that folds up to 50 open triage tickets from the same agent into one target ticket, marking the others merged, plus a case-insensitive search parameter on ticket lists that matches titles, descriptions and comments and returns a highlighted excerpt. Workflow End nodes now record an outcome of success or failure, new agents get call queueing enabled by default, and post-call audio webhooks can be retried. The same update adds optional audio effects such as reverb, distance, background noise and panning to Text to Speech and Text to Dialogue at no extra cost, and a turn-prediction commit strategy for realtime Speech to Text.
- ElevenLabs adds Brand Kit to ElevenCreative for logos, fonts and on-brand campaign generation
ElevenLabs says Brand Kit in ElevenCreative stores a logo, fonts and colors once so generated launch campaigns and other assets pull brand styles automatically.
- ElevenLabs adds synthetic voice detection in ElevenAgents and joins Personal Agent Protocol work
ElevenLabs is shipping synthetic voice detection in ElevenAgents for businesses facing AI-agent phone calls and says it is joining the emerging Personal Agent Protocol working group on how agents interact with companies.
- ElevenLabs partners with Banner Health on AI voice agents for primary care scheduling calls
ElevenLabs is partnering with Banner Health to answer patient calls with AI voice agents, starting with primary care scheduling. ElevenAgents books, reschedules or cancels appointments directly in Banner's electronic medical record at any hour, and hands calls to a Banner team member with context when a patient needs a person.
- ElevenLabs arrives on OpenRouter with nine text-to-speech and two speech-to-text models
OpenRouter now offers ElevenLabs' nine text-to-speech and two speech-to-text models, including Eleven v4, v4 Turbo and Scribe v2, through its speech and transcription endpoints with an OpenRouter API key and no separate ElevenLabs plan. Every ElevenLabs model is 50% off OpenRouter's list price through October 19, 8am PT.
- ElevenLabs' ConvAI widget core 0.19.0 adds end-call confirmation, carousel cards and rich content in voice calls
ElevenLabs released @elevenlabs/convai-widget-core 0.19.0 for its TypeScript Agents SDK. The minor release adds an optional confirmation dialog before ending a chat or call, enabled with the end-confirmation attribute or end_confirmation_enabled config and customizable copy; carousel rich content with scrollable item cards that can include an image, subtitle, description and actions, which stay in the transcript after a reply while their actions are removed; and rich content shown in voice calls when the transcript is visible. Patch changes update the dependency to @elevenlabs/client 1.27.0.
- ElevenLabs' TypeScript Agents SDK client 1.27.0 lets Scribe microphone transcription take prior text as context
ElevenLabs released @elevenlabs/client 1.27.0, part of its Agents SDK for TypeScript, which adds a previousText option to Scribe.connect() and useScribe() in microphone mode. The text is sent with the first microphone audio chunk to give the speech-to-text model context, such as existing document text, for casing, punctuation and sentence continuation. A patch change makes output resampling use a custom libsamplerate bundle path, matching the input side.
- ElevenLabs launches ElevenAgents Architect for team-wide agent management
ElevenLabs launched ElevenAgents Architect, which lets any member of a team manage agents wherever they work. Users can talk to it inside ElevenAgents to configure and maintain voice agents.
- Eleven v4 Turbo tops Artificial Analysis TTS arena at roughly 100 ms median latency
Eleven v4 Turbo has taken the top spot on Artificial Analysis' voice TTS arena, per ElevenLabs. The company cites a median inference latency of roughly 100 ms for the model, which launched in late September. Faster response is aimed at live voice-agent use cases.
- Eleven v4 voice model lands in ElevenReader
ElevenLabs brought its most expressive voice model, Eleven v4, into ElevenReader, per the company's announcement.