Gradium announced over 1,000 new AI voices covering English, French, Spanish, German, and Portuguese. The voices support customer service, narration, ads, and character use cases. They are available in 29 distinct accents, expanding the catalog from roughly 380 to about 1.4 k voices.
Key points
New voices: 1,000+ added to the catalog
Accents: 29 different accents across five languages
Catalog size: grew from ~380 to ~1.4 k voices
Use‑case distribution: 257 for customer service, 221 for narration, 228 for ads/social media, 329 for characters
Sources
OpenRouter now supports eleven TTS and STT models
OpenRouter is now live. You can use your existing OpenRouter API key to call eleven Text to Speech and Speech to Text models. Supported models include v4, v4 Turbo, and Scribe v2.
Key points
Launch: OpenRouter live for eleven models.
Models: v4, v4 Turbo, and Scribe v2 available.
Sources
Speech data collection scale and contributor details
The dataset is recorded by about 2.5 million contributors in roughly 180 countries. Each recording is paid and includes a consent record for the speaker. The accents reflect real user speech.
Key points
Contributors: ~2.5 M people worldwide
Coverage: recordings from ~180 countries
Sources
ElevenAgents Architect enters alpha
ElevenAgents Architect is now in alpha. The system lets agents flag problems, triage tickets, find root causes, fix code on a branch, test, and open a merge proposal. A human reviewer signs off before the fix is gradually deployed to users.
Key points
Self‑driven lifecycle: agents handle problem detection through merge proposal.
Human sign‑off triggers gradual rollout to users.
