GLM-TTS Evaluation Results
December 26, 2025 · View on GitHub
Model: glmtts (config)
Evaluation Date: 2025/12 (from res/glmtts/*/res-overall.json)
Paper/Repo: zai-org/GLM-TTS
Metrics Legend:
- WER⬇️: Word Error Rate (lower is better)
- CER⬇️: Character Error Rate (lower is better)
- SIM⬆️: Speaker Similarity (higher is better)
- P808_MOS⬆️: DNSMOS P.808 Mean Opinion Score (higher is better)
Seed-TTS-Eval Benchmark
| task | dataset | WER/CER⬇️ | SIM⬆️ | eval_cli | note |
|---|---|---|---|---|---|
| tts | seed_tts_eval_en | 2.48 | 67.25 | [1] | |
| tts | seed_tts_eval_zh | 1.06(1.03) | 75.98(76.1) | [2] |
CV3 Benchmark (Zero-Shot)
| task | dataset | WER/CER⬇️ | SIM⬆️ | P808_MOS⬆️ | eval_cli | note |
|---|---|---|---|---|---|---|
| tts | cv3_zero_shot_en | 6.61 | 72.08 | 3.70 | [3] | |
| tts | cv3_zero_shot_zh | 3.61 | 77.98 | 3.73 | [4] | |
| tts | cv3_zero_shot_hard_en | 9.40 | 73.43 | 3.73 | [5] | fail_rate: 3.12% |
| tts | cv3_zero_shot_hard_zh | 9.14 | 77.79 | 3.64 | [6] |
Evaluation Commands
[1] python audio_evals/main.py --dataset seed_tts_eval_en --model glmtts
[2] python audio_evals/main.py --dataset seed_tts_eval_zh --model glmtts
[3] python audio_evals/main.py --dataset cv3_zero_shot_en --model glmtts
[4] python audio_evals/main.py --dataset cv3_zero_shot_zh --model glmtts
[5] python audio_evals/main.py --dataset cv3_zero_shot_hard_en --model glmtts
[6] python audio_evals/main.py --dataset cv3_zero_shot_hard_zh --model glmtts