Voice Remixing live for Gemini 3.8 TTS via natural-language edits
On 4 Nov 2026 Google’s Gemini API release notes shipped Voice Remixing for Gemini 3.8 Flash TTS and Flash-Lite TTS. Developers can take a prebuilt or designed voice, apply natural-language instructions to adjust traits such as timbre, pitch, pace, and accent, and store a new independent voice—also available interactively in Google AI Studio.
Verified against Google’s Voice remixing guide and the Gemini API changelog entry for November 4, 2026, consistent with Google’s public note that the feature is live in Google AI Studio and the Gemini API:
What shipped
Voice remixing lets you modify an existing voice with natural-language instructions to create a new, independent stored voice, instead of designing a persona from scratch. A base_voice can be a prebuilt studio voice, an Extended Voice Library voice, or a custom Voice design persona. The input prompt describes relative changes (for example age feel, pitch, energy, timbre, or accent).
How to use it (official)
- Call
POST /v1beta/voiceswithgemini-3.8-flash-tts,typeprompted, plusbase_voiceand the edit prompt—or remix interactively in Google AI Studio’s speech UI. - You receive a new
voice_...ID and an optionalsample_audiopreview. - Synthesize with that ID on Gemini 3.8 Flash TTS (
gemini-3.8-flash-tts) or Gemini 3.8 Flash-Lite TTS (gemini-3.8-flash-lite-tts).
Docs show remixing prebuilt Puck into an older, warmer British audiobook-style voice, then synthesizing with the new ID. Google’s public post also notes starting from a 2,000+ voice library or a custom-designed voice and fine-tuning timbre, pitch, pace, and accent with prompts (for example “add a subtle Southern U.S. accent”).
This summary follows the official docs and changelog; it does not invent pricing or quota changes not stated there.