Skip to content

Benchmark

EmergentTTS-Eval WER

Scores are percentages. Lower is better.

Models measured
9
Scores recorded
14
Sources
1
IQ weight
0.81

Score distribution

12.3%
Mean
11.9%
Median
2.2%
Std dev
10.3%
10th percentile
16.5%
90th percentile
10.1%
Min
10.1%17.7%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 0.81 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
1Speech 02 HDvoice: English_expressive_narrator· EmergentTTS-Eval data10.1%
2Gemini 2.5 Flash TTSvoice: Zephyr· EmergentTTS-Eval data10.3%
3Gemini 2.5 Flash TTSvoice: Zephyr, strong-prompt· EmergentTTS-Eval data10.4%
4GPT-4o mini TTSvoice: Alloy, baseline· EmergentTTS-Eval data10.6%
5GPT-4o mini TTSvoice: Alloy, strong-prompt· EmergentTTS-Eval data10.8%
6GPT-4o mini Audio Previewvoice: Alloy· EmergentTTS-Eval data10.9%
7Gemini 2.5 Pro TTSvoice: Zephyr, strong-prompt· EmergentTTS-Eval data11.8%
8GPT-4o Audio Previewvoice: Ballad, strong-prompt· EmergentTTS-Eval data11.9%
9GPT-4o Audio Previewvoice: Alloy, strong-prompt· EmergentTTS-Eval data12%
10GPT-4o Audio Previewvoice: Alloy· EmergentTTS-Eval data12.4%
11GPT-4o mini Audio Previewvoice: Alloy, strong-prompt· EmergentTTS-Eval data13.1%
12Kokoro 82Mvoice: af_heart· EmergentTTS-Eval data13.4%
13F5 TTSvoice: basic_ref_en· EmergentTTS-Eval data16.5%
14Orpheus TTSvoice: Tara· EmergentTTS-Eval data17.7%

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