Today for AI
HOT RADAR
llmHEAT 14.6°

AI CURATED HOT STORY · 10/31/2026

Voice Arena Launches Turing TTS Benchmark; Top Models Fool Listeners Only 10%

1 reports archived1 independent sourcesupdated 10/31/2026, 03:06:17
Synthesis & Latest Updates
1 Sources Cross-Validated

Voice Arena introduced the Turing TTS benchmark to decouple human-likeness from user preference in voice AI evaluation. Results show that even the top-performing model, Gemini 3.8 Flash TTS, fooled listeners only ~10% of the time, indicating a significant gap remains between current AI voices and real human speech.

LATEST/Voice Arena introduced the Turing TTS benchmark to decouple human-likeness from user preference in voice AI evaluation. Results show that even the top-performing model, Gemini 3.8 Flash TTS, fooled listeners only ~10% of the time, indicating a significant gap remains between current AI voices and real human speech.

Hourly heat curve

2 fully observed hours
Fewer than 3 fully observed hours; trend not plotted yet.

Coverage timeline

Total 1 reports · Latest first
  1. Rohan Paul (@rohanpaul_ai)T2·68 pts

    Voice Arena Launches Turing TTS Benchmark; Top Models Fool Listeners Only 10%

    • Turing TTS scores 'human-likeness' directly by having listeners distinguish between real humans and models, distinct from traditional preference rankings.
    • The best model, Gemini 3.8 Flash TTS, achieved a 'Fool Rate' of only ~10%, far below the 50% indistinguishability threshold.
    • While the top four models are statistically tied on Voice Arena's preference board, the entire field still has a long way to go regarding realism.