Benchmark
EmergentTTS-Eval Complex Pronunciation WER
Scores are percentages. Lower is better.
- Models measured
- 9
- Scores recorded
- 14
- Sources
- 1
- IQ weight
- 0.75
Score distribution
- 34.4%
- Mean
- 34.8%
- Median
- 3.9%
- Std dev
- 29.9%
- 10th percentile
- 41%
- 90th percentile
- 28.8%
- Min
This benchmark has a weight of 0.75 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
| # | Model | Score |
|---|---|---|
| 1 | Speech 02 HDvoice: English_expressive_narrator· EmergentTTS-Eval data | 28.8% |
| 2 | GPT-4o mini TTSvoice: Alloy, baseline· EmergentTTS-Eval data | 29.9% |
| 3 | Kokoro 82Mvoice: af_heart· EmergentTTS-Eval data | 30.1% |
| 4 | GPT-4o mini TTSvoice: Alloy, strong-prompt· EmergentTTS-Eval data | 31.6% |
| 5 | Gemini 2.5 Flash TTSvoice: Zephyr· EmergentTTS-Eval data | 31.9% |
| 6 | GPT-4o mini Audio Previewvoice: Alloy· EmergentTTS-Eval data | 32.3% |
| 7 | Gemini 2.5 Flash TTSvoice: Zephyr, strong-prompt· EmergentTTS-Eval data | 33.6% |
| 8 | Gemini 2.5 Pro TTSvoice: Zephyr, strong-prompt· EmergentTTS-Eval data | 34.8% |
| 9 | GPT-4o Audio Previewvoice: Ballad, strong-prompt· EmergentTTS-Eval data | 35.3% |
| 10 | GPT-4o Audio Previewvoice: Alloy· EmergentTTS-Eval data | 35.9% |
| 11 | GPT-4o Audio Previewvoice: Alloy, strong-prompt· EmergentTTS-Eval data | 36.9% |
| 12 | GPT-4o mini Audio Previewvoice: Alloy, strong-prompt· EmergentTTS-Eval data | 37.1% |
| 13 | Orpheus TTSvoice: Tara· EmergentTTS-Eval data | 41% |
| 14 | F5 TTSvoice: basic_ref_en· EmergentTTS-Eval data | 42.4% |
Data credits: EmergentTTS-Eval (Boson AI), Apache-2.0 (licence)