通过增强对抗性流匹配，提升语音合成对错位的鲁棒性。
RobustSpeechFlow: Learning Robust Text-to-Speech Trajectories via Augmentation-based Contrastive Flow Matching
- 引入保持长度的重复与跳过增强，直接惩罚真实错误模式。
- 在零样本场景下，英文字符错误率从0.48%降至0.35%。
- 无需外部对齐器，可无缝集成到现有语音合成流程中。
尽管基于流匹配的文本到语音（TTS）系统在零样本说话人相似性和自然度方面表现优异，但仍面临内容保真度问题，尤其在对齐不完美时易出现跳词和重复错误。本文提出RobustSpeechFlow，一种训练策略，通过在对比流匹配中引入保持长度的重复与跳词潜在增强，提升对齐鲁棒性。该方法无需外部对齐器或偏好数据，直接惩罚真实故障模式，并可轻松融入现有流水线。在Seed-TTS-eval上，仅使用0.06B参数，将词错误率（WER）从1.44降至1.38；在ZERO500基准测试中，跨不同说话人和语调条件均实现一致可懂度提升：在NFE=24时，英文字符错误率（CER）由0.48%降至0.35%，韩文CER由0.81%降至0.57%。
原文摘要 · Abstract (English)
While flow-matching text-to-speech (TTS) achieves strong zero-shot speaker similarity and naturalness, it remains susceptible to content fidelity issues, particularly skip and repeat errors from imperfect alignment. We propose RobustSpeechFlow, a training strategy that improves alignment robustness by extending contrastive flow matching with length-preserving repeat and skip latent augmentations. Requiring no external aligners or preference data, our method directly penalizes realistic failure modes and readily integrates into existing pipelines. On Seed-TTS-eval, it reduces the word error rate (WER) from 1.44 to 1.38 using only 0.06B parameters. On our ZERO500 benchmark, it delivers consistent intelligibility improvements across diverse speaker and prosody conditions; at NFE=24, it reduces English character error rate (CER) from 0.48\% to 0.35\% and Korean CER from 0.81\% to 0.57\%. Audio samples: https://robustspeechflow.github.io/
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。