Skip to content

Benchmark

EmergentTTS-Eval Emotions Win-Rate

Scores are percentages. Higher is better.

Models measured
9
Scores recorded
13
Sources
1
IQ weight
0.34

Score distribution

64.5%
Mean
67.1%
Median
25.6%
Std dev
29.6%
10th percentile
97%
90th percentile
18%
Min
18%97.3%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 0.34 for speech models, measured over 7 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.

Leaderboard

#ModelScore
1Gemini 2.5 Flash TTSvoice: Zephyr, strong-prompt· EmergentTTS-Eval data97.3%
2GPT-4o Audio Previewvoice: Ballad, strong-prompt· EmergentTTS-Eval data97%
3Gemini 2.5 Flash TTSvoice: Zephyr· EmergentTTS-Eval data95%
4Gemini 2.5 Pro TTSvoice: Zephyr, strong-prompt· EmergentTTS-Eval data92.1%
5GPT-4o mini Audio Previewvoice: Alloy, strong-prompt· EmergentTTS-Eval data74.1%
6GPT-4o mini TTSvoice: Alloy, strong-prompt· EmergentTTS-Eval data67.3%
7GPT-4o Audio Previewvoice: Alloy, strong-prompt· EmergentTTS-Eval data67.1%
8GPT-4o mini Audio Previewvoice: Alloy· EmergentTTS-Eval data62.9%
9GPT-4o Audio Previewvoice: Alloy· EmergentTTS-Eval data56.8%
10Speech 02 HDvoice: English_expressive_narrator· EmergentTTS-Eval data41.6%
11Orpheus TTSvoice: Tara· EmergentTTS-Eval data39.1%
12F5 TTSvoice: basic_ref_en· EmergentTTS-Eval data29.6%
13Kokoro 82Mvoice: af_heart· EmergentTTS-Eval data18%

Data credits: EmergentTTS-Eval (Boson AI), Apache-2.0 (licence)