CL-SN55 941 —
GPT-61S 890 —
GPT-56T 861 —
GPT-6S 837 -1.2%
CL-OP55X 831 -1.4%
MUSE-SPK 804 —
QWEN-38X 795 —
GPT-6A 782 —
CL-OP5H 724 -1.6%
CL-OP5X 722 -1.6%
KIMI-K3X 718 -0.1%
CL-OP46H 707 -1.7%
CL-FAB5H 705 -7.4%
CL-OP55H 705 -1.7%
CL-OP47H 696 -1.7%
GEM-37FH 661 -9.7%
GEM-38FH 656 -4.8%
GPT-56S 578 -0.5%
CL-OP47 543 -0.5%
INKL 531 —
GEM-3P 494 -0.2%
CL-OP46 490 -0.2%
GEM-31P 485 -0.4%
CL-OP48 483 -0.2%
GPT-6L 436 —
CL-SN55 941 —
GPT-61S 890 —
GPT-56T 861 —
GPT-6S 837 -1.2%
CL-OP55X 831 -1.4%
MUSE-SPK 804 —
QWEN-38X 795 —
GPT-6A 782 —
CL-OP5H 724 -1.6%
CL-OP5X 722 -1.6%
KIMI-K3X 718 -0.1%
CL-OP46H 707 -1.7%
CL-FAB5H 705 -7.4%
CL-OP55H 705 -1.7%
CL-OP47H 696 -1.7%
GEM-37FH 661 -9.7%
GEM-38FH 656 -4.8%
GPT-56S 578 -0.5%
CL-OP47 543 -0.5%
INKL 531 —
GEM-3P 494 -0.2%
CL-OP46 490 -0.2%
GEM-31P 485 -0.4%
CL-OP48 483 -0.2%
GPT-6L 436 —
← Back to feed

Gemini 3.8 Flash TTS Claims #1 on Hume AI Voice Benchmark with 71.4, Adds 2,000-Voice Library

Google launched two new text-to-speech models on September 23: Gemini 3.8 Flash TTS and Gemini 3.8 Flash-Lite TTS. Both are live now via Gemini API and Google AI Studio.

Benchmark position

  • Hume AI Voice Design Benchmark: 71.4 (#1 overall)
  • Accent modeling: 60.8 (#1)
  • Hume AI Overall Quality Index: #1 (Flash TTS) and #2 (Flash-Lite TTS)
  • Voice Arena: top positions in Japanese, Brazilian Portuguese, Vietnamese, MSA Arabic, Mexican Spanish, and Hindi

Gemini 3.1 Flash TTS launched in April 2026 with 30 preset voices and 70+ language support. The 3.8 generation replaces that preset model with generative voice design.

What changed from 3.1

  • Generative voice design: describe a character in natural language and the model creates a matching voice from scratch, across 100+ languages and dialects
  • Voice library: 2,000+ production-ready voices, including regional variants such as Mexican Spanish, Quebec French, and Scots English
  • Voice replication: reconstruct a consistent vocal profile from a 30-second audio sample, backed by SynthID watermarking and C2PA credentials for consent verification
  • Line-level direction: control acting cues, pacing, dialect shifts, and backchanneling on individual lines of a script

Flash-Lite TTS

The lighter model targets high-volume pipelines: dubbing, voice agent infrastructure, and content creation at scale. Google positions it as cost-efficient rather than creatively expressive, with fine-grained tone and pacing controls but without the full generative voice-design stack.

Availability

  • Developers: Gemini API and Google AI Studio (live now)
  • Enterprises: Gemini Enterprise API (coming soon)
  • Products: Gemini Notebook and Google Vids (shipping with the new models)

The 3.8 TTS pair extends Google’s audio push that began with Gemini 3.5 Live Translate, 3.5 Transcribe, and 3.8 Live earlier this year. Voice generation is now a direct competitive front between labs alongside coding and reasoning.