Release Emilia and Emilia-Pipe

[yuantuo666]

✨ Description

We release the Emilia, an extensive, multilingual, and diverse dataset, and Emilia-Pipe, the first open-source preprocessing pipeline designed to transform in-the-wild speech data into high-quality training data with annotations for speech generation.

Major contribution for this PR: @HarryHe11 @shangqwe123 @yuantuo666 @lixuyuan102

🚧 Related Issues

None

👨‍💻 Changes Proposed

• Update README.md News
• Add a README.md in preprocessors/Emilia directory to introduce Emilia-Pipe
• Integrate processing pipeline Emilia-Pipe under preprocessors/Emilia directory

🧑‍🤝‍🧑 Who Can Review?

@HarryHe11 @jiaqili3 @RMSnow @HeCheng0625

🛠 TODO

None

✅ Checklist

• Code has been reviewed
• Code complies with the project's code standards and best practices
• Code has passed all tests
• Code does not affect the normal use of existing features
• Code has been commented properly
• Documentation has been updated (if applicable)
• Demo/checkpoint has been attached (if applicable)

[HarryHe11]

Hi Chaoren @yuantuo666 , thank you so much for raising this PR and all of your efforts @yuantuo666 @shangqwe123 @lixuyuan102!

I notice that we haven't provide the list for the source audios yet in this pr. Maybe we could complete this part before merging!

[HarryHe11]

@yuantuo666 @lixuyuan102 @shangqwe123

Thanks so much for Chaoren's efforts in preparing this final version of Emilia-Pipe, @yuantuo666! I think most of the code is of very good quality, neat, and well-commented!

Before merging this PR, I think we should provide the lists for the source audios and address my comments, as well as the important comments from Xueyao, Jiaqi, and Yuancheng. @RMSnow @jiaqili3 @HeCheng0625

[RMSnow]

Great works! The most important thing is to add some high-level descriptions for Emilia. Such a description should appear both in Emilia's readme and Amphion's main readme.

[HarryHe11]

@RMSnow Hi Xueyao, thank you so much for your helpful suggestions! I have addressed all of your comments accordingly!

[HarryHe11]

@yuantuo666 @shangqwe123 @lixuyuan102 Chaoren, Zengqiang, Xuyuan,

I think we'd better merge this pr after we got our arxiv link, and done all of the following tasks.

• Codes are up to date (all tasks are completed)
• Codes are tested.
• Demopage is up to date (the abstract is consistent with our arxiv paper).
• Licenses and copyright declarations have been appropriately added in Github, huggingface.
• Huggingface releases the meta-information for the source audios.
• Amphion's main README relayed the news of Emilia's release and provides a link to the sub-repo.
• Emilia's README introduces the dataset adequately.
• Emilia's README provides appropriate URLs to the demopage. 2. Hugging Face Page. 3. Arxiv paper
• Emilia's README, Demopage, and huggingface display references to Emilia's arxiv work and amphion.

[RMSnow]

Great works!