OpenVoice:多功能即时声音克隆技术

OpenVoice is a multi-functional real-time voice cloning technology developed by the MyShell team. It can clone the voice of the original speaker and generate speech in various languages by providing a short audio sample.

It has the following advantages:

  1. High-precision voice cloning:
    It can highly restore the reference voice, supporting speech generation in multiple languages and accents.

  2. Flexible voice style control:
    It allows fine adjustments to the emotions and accents of the voice, as well as control over rhythm, pauses, and intonation.

  3. Cross-language voice cloning without sample:
    Both the reference voice and generated voice can be in any language outside of large-scale multilingual datasets.

GitHub: github.com/myshell-ai/OpenVoice

I tested it, and the effect for Chinese is not very good. The MyShell team is aware of this issue and has expressed that they are working on optimizing it.
OpenVoice:多功能即时声音克隆技术。

这是由 MyShell 团队开发的一项技术,只需提供一段简短的音频样本,就能克隆出原发言者的声音,并能以此生成各种语言的语音。

它具有如下优势:

1)高精度音色克隆
能够高度还原参考音色,支持多语种和多种口音的语音生成。

2)灵活的声音风格调控
可以对声音的情感、口音进行精细调整,还可以控制节奏、停顿和语调等多种声音风格。

3)无需样本的跨语言声音克隆
无论是参考声音还是生成的声音,都可以是大型多语种数据集之外的任何语言。

GitHub:github.com/myshell-ai/OpenVoice

测试了下,中文效果不是很好, 对此 MyShell 团队也清楚并表示正在优化处理。

image.png