CosyVoice:基于有监督语义标记的可扩展多语言零镜头文本到语音合成器

Zhihao Du, Qian Chen, Shiliang Zhang, Kai Hu, Heng Lu, Yexin Yang, Hangrui Hu, Siqi Zheng, Yue Gu, Ziyang Ma, Zhijie Yan
{"title":"CosyVoice:基于有监督语义标记的可扩展多语言零镜头文本到语音合成器","authors":"Zhihao Du, Qian Chen, Shiliang Zhang, Kai Hu, Heng Lu, Yexin Yang, Hangrui Hu, Siqi Zheng, Yue Gu, Ziyang Ma, Zhijie Yan","doi":"arxiv-2407.05407","DOIUrl":null,"url":null,"abstract":"Recent years have witnessed a trend that large language model (LLM) based\ntext-to-speech (TTS) emerges into the mainstream due to their high naturalness\nand zero-shot capacity. In this paradigm, speech signals are discretized into\ntoken sequences, which are modeled by an LLM with text as prompts and\nreconstructed by a token-based vocoder to waveforms. Obviously, speech tokens\nplay a critical role in LLM-based TTS models. Current speech tokens are learned\nin an unsupervised manner, which lacks explicit semantic information and\nalignment to the text. In this paper, we propose to represent speech with\nsupervised semantic tokens, which are derived from a multilingual speech\nrecognition model by inserting vector quantization into the encoder. Based on\nthe tokens, we further propose a scalable zero-shot TTS synthesizer, CosyVoice,\nwhich consists of an LLM for text-to-token generation and a conditional flow\nmatching model for token-to-speech synthesis. Experimental results show that\nsupervised semantic tokens significantly outperform existing unsupervised\ntokens in terms of content consistency and speaker similarity for zero-shot\nvoice cloning. Moreover, we find that utilizing large-scale data further\nimproves the synthesis performance, indicating the scalable capacity of\nCosyVoice. To the best of our knowledge, this is the first attempt to involve\nsupervised speech tokens into TTS models.","PeriodicalId":501178,"journal":{"name":"arXiv - CS - Sound","volume":null,"pages":null},"PeriodicalIF":0.0000,"publicationDate":"2024-07-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"CosyVoice: A Scalable Multilingual Zero-shot Text-to-speech Synthesizer based on Supervised Semantic Tokens\",\"authors\":\"Zhihao Du, Qian Chen, Shiliang Zhang, Kai Hu, Heng Lu, Yexin Yang, Hangrui Hu, Siqi Zheng, Yue Gu, Ziyang Ma, Zhijie Yan\",\"doi\":\"arxiv-2407.05407\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"Recent years have witnessed a trend that large language model (LLM) based\\ntext-to-speech (TTS) emerges into the mainstream due to their high naturalness\\nand zero-shot capacity. In this paradigm, speech signals are discretized into\\ntoken sequences, which are modeled by an LLM with text as prompts and\\nreconstructed by a token-based vocoder to waveforms. Obviously, speech tokens\\nplay a critical role in LLM-based TTS models. Current speech tokens are learned\\nin an unsupervised manner, which lacks explicit semantic information and\\nalignment to the text. In this paper, we propose to represent speech with\\nsupervised semantic tokens, which are derived from a multilingual speech\\nrecognition model by inserting vector quantization into the encoder. Based on\\nthe tokens, we further propose a scalable zero-shot TTS synthesizer, CosyVoice,\\nwhich consists of an LLM for text-to-token generation and a conditional flow\\nmatching model for token-to-speech synthesis. Experimental results show that\\nsupervised semantic tokens significantly outperform existing unsupervised\\ntokens in terms of content consistency and speaker similarity for zero-shot\\nvoice cloning. Moreover, we find that utilizing large-scale data further\\nimproves the synthesis performance, indicating the scalable capacity of\\nCosyVoice. To the best of our knowledge, this is the first attempt to involve\\nsupervised speech tokens into TTS models.\",\"PeriodicalId\":501178,\"journal\":{\"name\":\"arXiv - CS - Sound\",\"volume\":null,\"pages\":null},\"PeriodicalIF\":0.0000,\"publicationDate\":\"2024-07-07\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"arXiv - CS - Sound\",\"FirstCategoryId\":\"1085\",\"ListUrlMain\":\"https://doi.org/arxiv-2407.05407\",\"RegionNum\":0,\"RegionCategory\":null,\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"\",\"JCRName\":\"\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"arXiv - CS - Sound","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/arxiv-2407.05407","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
引用次数: 0

摘要

近年来,基于大语言模型(LLM)的文本到语音(TTS)因其高自然度和零误差能力而成为主流趋势。在这种模式中,语音信号被离散化为令牌序列,由 LLM 以文本作为提示进行建模,并由基于令牌的声码器将其重组为波形。显然,语音标记在基于 LLM 的 TTS 模型中起着至关重要的作用。目前的语音标记是以无监督的方式学习的,缺乏明确的语义信息和与文本的对齐。在本文中,我们提出用有监督的语义标记来表示语音,这种标记来自多语言语音识别模型,通过在编码器中插入向量量化来实现。在语义标记的基础上,我们进一步提出了一种可扩展的 "0-shot TTS "合成器--CosyVoice,它由用于文本到标记生成的 LLM 和用于标记到语音合成的条件流匹配模型组成。实验结果表明,在内容一致性和说话人相似性方面,有监督语义标记在零点语音克隆方面明显优于现有的无监督标记。此外,我们还发现利用大规模数据可以进一步提高合成性能,这表明 CosyVoice 具有可扩展性。据我们所知,这是首次尝试将有监督的语音标记纳入 TTS 模型。
本文章由计算机程序翻译,如有差异,请以英文原文为准。
CosyVoice: A Scalable Multilingual Zero-shot Text-to-speech Synthesizer based on Supervised Semantic Tokens
Recent years have witnessed a trend that large language model (LLM) based text-to-speech (TTS) emerges into the mainstream due to their high naturalness and zero-shot capacity. In this paradigm, speech signals are discretized into token sequences, which are modeled by an LLM with text as prompts and reconstructed by a token-based vocoder to waveforms. Obviously, speech tokens play a critical role in LLM-based TTS models. Current speech tokens are learned in an unsupervised manner, which lacks explicit semantic information and alignment to the text. In this paper, we propose to represent speech with supervised semantic tokens, which are derived from a multilingual speech recognition model by inserting vector quantization into the encoder. Based on the tokens, we further propose a scalable zero-shot TTS synthesizer, CosyVoice, which consists of an LLM for text-to-token generation and a conditional flow matching model for token-to-speech synthesis. Experimental results show that supervised semantic tokens significantly outperform existing unsupervised tokens in terms of content consistency and speaker similarity for zero-shot voice cloning. Moreover, we find that utilizing large-scale data further improves the synthesis performance, indicating the scalable capacity of CosyVoice. To the best of our knowledge, this is the first attempt to involve supervised speech tokens into TTS models.
求助全文
通过发布文献求助,成功后即可免费获取论文全文。 去求助
来源期刊
自引率
0.00%
发文量
0
×
引用
GB/T 7714-2015
复制
MLA
复制
APA
复制
导出至
BibTeX EndNote RefMan NoteFirst NoteExpress
×
提示
您的信息不完整,为了账户安全,请先补充。
现在去补充
×
提示
您因"违规操作"
具体请查看互助需知
我知道了
×
提示
确定
请完成安全验证×
copy
已复制链接
快去分享给好友吧!
我知道了
右上角分享
点击右上角分享
0
联系我们:info@booksci.cn Book学术提供免费学术资源搜索服务,方便国内外学者检索中英文文献。致力于提供最便捷和优质的服务体验。 Copyright © 2023 布克学术 All rights reserved.
京ICP备2023020795号-1
ghs 京公网安备 11010802042870号
Book学术文献互助
Book学术文献互助群
群 号:481959085
Book学术官方微信