--- format: "aidr-story-markdown/v1" id: "e5d1d3d211c0e4014cd5367181362b4557cee4067a8af2c66838ea9615d0137e" canonical_url: "https://aidr.today/e5d1d3d2?lang=en" title: "Google Gemini 3.8 Flash TTS Takes #1 in Pronunciation Robustness" lang: "en" requested_lang: "en" available_langs: ["en","vi"] translation_fallback: null fallback_fields: [] published_at: "2026-09-23T17:17:53.000Z" category: "Models" topics: ["google","gemini","text-to-speech","multimodal","deepmind","anthropic","claude","multi-agent"] source_urls: ["https://marketbrief.now/ai/google-gemini-38-flash-tts-takes-1-in-pronunciation-robustness-345dad37","https://huggingnews.com/ai/google-gemini-38-flash-tts-takes-1-in-pronunciation-robustness-345dad37","https://huggingnews.com/ai/google-gemini-38-flash-tts-hits-1-in-pronunciation-robustness-at-launch-8c6b678c","https://marketbrief.now/ai/google-gemini-38-flash-tts-hits-1-in-pronunciation-robustness-at-launch-8c6b678c"] summary: "Google DeepMind released Gemini 3.8 Flash TTS and Gemini 3.8 Flash-Lite TTS, two text-to-speech models featuring voice replication and support for 100 languages. The Flash TTS model scored 89.5% on the Pronunciation Robustness Benchmark, taking the top spot from the previous leader, Gemini 3.1 Flash TTS. It also debuted at #2 on the Provider Voice Arena Leaderboard with an Elo of 1,263, while the Flash-Lite version ranked #6 with 1,236. The models are available via the Gemini API and AI Studio with a library of over 2,000 voices. Gemini 3.8 Flash TTS costs $32.98 per 1 million characters, and Flash-Lite is $22.07 per 1 million characters, both priced higher than the $18.31 rate for Gemini 3.1 Flash TTS but below Eleven v3 at $100 per 1 million characters. In terms of speed, the Flash model processes 44.1 characters per second, roughly 2.7 times faster than real-time. However, it trails other competitors such as Falcon 2, which processes 204.9 characters per second, and Luna TTS at 153.5 characters per second." --- # Google Gemini 3\.8 Flash TTS Takes \#1 in Pronunciation Robustness > [Open the canonical story]() **Published:** 2026-09-23T17:17:53.000Z **Category:** Models **Topics:** google, gemini, text\-to\-speech, multimodal, deepmind, anthropic, claude, multi\-agent ## Summary Google DeepMind released Gemini 3\.8 Flash TTS and Gemini 3\.8 Flash\-Lite TTS, two text\-to\-speech models featuring voice replication and support for 100 languages\. The Flash TTS model scored 89\.5% on the Pronunciation Robustness Benchmark, taking the top spot from the previous leader, Gemini 3\.1 Flash TTS\. It also debuted at \#2 on the Provider Voice Arena Leaderboard with an Elo of 1,263, while the Flash\-Lite version ranked \#6 with 1,236\. The models are available via the Gemini API and AI Studio with a library of over 2,000 voices\. Gemini 3\.8 Flash TTS costs $32\.98 per 1 million characters, and Flash\-Lite is $22\.07 per 1 million characters, both priced higher than the $18\.31 rate for Gemini 3\.1 Flash TTS but below Eleven v3 at $100 per 1 million characters\. In terms of speed, the Flash model processes 44\.1 characters per second, roughly 2\.7 times faster than real\-time\. However, it trails other competitors such as Falcon 2, which processes 204\.9 characters per second, and Luna TTS at 153\.5 characters per second\. ## Sources - [Story source]() - [Story source]() - [Story source]() - [Story source]()