Skip to content

Benchmark

EmergentTTS-Eval Syntactic Complexity Win-Rate

Scores are percentages. Higher is better.

Models measured
9
Scores recorded
13
Sources
1
IQ weight
0.43

Score distribution

53.7%
Mean
58.9%
Median
13.4%
Std dev
29.5%
10th percentile
68%
90th percentile
25%
Min
25%68.4%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 0.43 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
1GPT-4o Audio Previewvoice: Ballad, strong-prompt· EmergentTTS-Eval data68.4%
2Gemini 2.5 Flash TTSvoice: Zephyr· EmergentTTS-Eval data68%
3Gemini 2.5 Flash TTSvoice: Zephyr, strong-prompt· EmergentTTS-Eval data67.1%
4GPT-4o mini TTSvoice: Alloy, strong-prompt· EmergentTTS-Eval data62%
5GPT-4o mini Audio Previewvoice: Alloy, strong-prompt· EmergentTTS-Eval data60.4%
6Gemini 2.5 Pro TTSvoice: Zephyr, strong-prompt· EmergentTTS-Eval data60%
7GPT-4o Audio Previewvoice: Alloy, strong-prompt· EmergentTTS-Eval data58.9%
8GPT-4o Audio Previewvoice: Alloy· EmergentTTS-Eval data56.8%
9GPT-4o mini Audio Previewvoice: Alloy· EmergentTTS-Eval data50.2%
10Kokoro 82Mvoice: af_heart· EmergentTTS-Eval data46.8%
11Speech 02 HDvoice: English_expressive_narrator· EmergentTTS-Eval data45.5%
12Orpheus TTSvoice: Tara· EmergentTTS-Eval data29.5%
13F5 TTSvoice: basic_ref_en· EmergentTTS-Eval data25%

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