Google adds prompt-based voice remixing to Gemini 3.8 Flash TTS

Reshape any of 2,000+ voices by prompt into a new reusable voice ID — plus Claude reportedly gets live AP election data, and an EU adviser warns on Gemini.

Nowline NOV 5 7:00 PM banner

Top AI stories from the last hour

Top AI stories from the last hour

Copy markdown

  • Reshape a voice with one prompt

    Take any of 2,000+ preset or custom-designed voices in Gemini 3.8 Flash TTS and adjust timbre, pitch, pace, and accent with plain English — "add a subtle Southern U.S. accent" — then save the result as an independent, reusable voice via POST /v1beta/voices on gemini-3.8-flash-tts. It builds on September's voice design and cloning; remixing is the new piece, live now in the API and AI Studio. For voice-agent builders it means dialing in a brand voice without recording fresh samples or retraining a thing.

  • Claude reportedly gets live AP election data

    ADWEEK reports Anthropic struck its first paid publisher licensing deal — with the Associated Press — so Claude served attributed, up-to-date AP figures on U.S. election night. Scope (consumer app vs API) and terms aren't confirmed, and ADWEEK is the sole origin. Still, it signals Anthropic warming to content licensing after holding out longer than OpenAI and Google, which could widen what Claude is allowed to cite in real time.

  • EU adviser: Gemini can infringe press rights

    An EU Court of Justice advocate general said Gemini's answers may infringe publishers' rights two ways — reproducing protected expression and circumventing paywalls (Case C-250/25, Like Company v Google). The opinion is non-binding, but AG opinions tend to preview the final ruling. If it holds, anyone building answer engines or RAG pipelines over news content in the EU faces a new liability line to design around.