← Azimuth Daily
Capturing data…
Capturing data…
FRI, 25 SEPT · 76 ITEMS
"Google releases Gemini 3.8 text-to-speech" — Hacker News (best) · AI & Technology
Google has released two new text-to-speech models under the Gemini 3.8 family: Flash and TTS. These models are designed for expressive, studio-quality audio generation, supporting prompt-based custom voice design, over 2,000 preset voices, and more than 100 languages. The release marks a significant step in making AI-generated speech more controllable and natural-sounding for applications like character voices and creative audio production.
The claim is solid and directly sourced from Google's official blog post and developer documentation, both published within the last day. The reporting closely mirrors the primary source, with additional context from developer-focused sites and social media confirming the same details.
This is genuinely new: the models and their announcement are from the past 24 hours, with no indication of being a resurfacing of older material.
We covered Google's release of Gemini-3.5-Transcribe and Gemini Omni 1.1 Flash in August 2026, which focused on transcription and multimodal capabilities, not text-to-speech. The current release is a distinct new product line within the Gemini model family.