Skip to content

Long-TTS-Eval.md

Latest commit

 

History

History
31 lines (22 loc) · 1.36 KB

Long-TTS-Eval.md

File metadata and controls

31 lines (22 loc) · 1.36 KB

Long-TTS-Eval Benchmark Results

Benchmark: Long-TTS-Eval Dataset | Task Config Evaluation Date: 2025/11/27 Paper: Long-TTS-Eval: A Benchmark for Evaluating Long-form Text-to-Speech

Metrics Legend:

  • WER⬇️: Word Error Rate (lower is better)
  • CER⬇️: Character Error Rate (lower is better)

Long-TTS-Eval Standard Set

Note: Performance format: reproduced_result(official_result) - values in parentheses are official results from the paper.

task sub measure performance eval_cli note
tts long_tts_eval_zh cer⬇️ 7.23(5.58) [1]
tts long_tts_eval_en wer⬇️ 4.69(4.98) [2]
tts long_tts_eval_hard_zh cer⬇️ 24.33(23.58) [3] JIA-Lab-research/MGM-Omni#6
tts long_tts_eval_hard_en wer⬇️ 32.84(26.26) [4]

Evaluation Commands

[1] python audio_evals/main.py --dataset long_tts_eval_zh --model mgm-omni-tts-zh [2] python audio_evals/main.py --dataset long_tts_eval_en --model mgm-omni-tts [3] python audio_evals/main.py --dataset long_tts_eval_hard_zh --model mgm-omni-tts-zh [4] python audio_evals/main.py --dataset long_tts_eval_hard_en --model mgm-omni-tts