arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2402.01692cs.CLcs.LG
收藏

通过自监督表示混合与嵌入初始化最大化跨语言TTS适应的数据效率

Maximizing Data Efficiency for Cross-Lingual TTS Adaptation by Self-Supervised Representation Mixing and Embedding Initialization

National Taiwan University(国立台湾大学)

Wei-Ping Huang, Sung-Feng Huang, Hung-yi Lee

AI总结:

本文提出一种数据高效的跨语言TTS适应框架,通过自监督特征混合和嵌入初始化,仅用4条标记数据和15分钟未标记数据即可合成清晰语音,并超越传统方法。

摘要:

本文提出了一种有效的迁移学习框架,用于文本到语音系统中的语言适应,重点在于使用最少的标记和未标记数据实现语言适应。尽管许多工作致力于减少标记数据的使用,但很少有研究考虑最小化未标记数据的使用。通过在预训练阶段利用自监督特征,在微调过程中用这些特征替换伪标签的噪声部分,并结合嵌入初始化技巧,我们的方法相比传统方法从未标记数据中利用了更多信息。实验结果表明,我们的框架仅使用4条标记数据和15分钟未标记数据就能合成未见语言的清晰语音。即使当可获取的数据量更大时,我们的方法仍持续超越传统技术。这些发现凸显了我们数据高效语言适应框架的潜力。

英文摘要:

This paper presents an effective transfer learning framework for language adaptation in text-to-speech systems, with a focus on achieving language adaptation using minimal labeled and unlabeled data. While many works focus on reducing the usage of labeled data, very few consider minimizing the usage of unlabeled data. By utilizing self-supervised features in the pretraining stage, replacing the noisy portion of pseudo labels with these features during fine-tuning, and incorporating an embedding initialization trick, our method leverages more information from unlabeled data compared to conventional approaches. Experimental results show that our framework is able to synthesize intelligible speech in unseen languages with only 4 utterances of labeled data and 15 minutes of unlabeled data. Our methodology continues to surpass conventional techniques, even when a greater volume of data is accessible. These findings highlight the potential of our data-efficient language adaptation framework.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑