Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Add sgmse implementation #177

Open
wants to merge 3 commits into
base: main
Choose a base branch
from
Open

Add sgmse implementation #177

wants to merge 3 commits into from

Conversation

lithr1
Copy link

@lithr1 lithr1 commented Apr 7, 2024

You can see the egs/sgmse/README.md. This task is for the final project for AIR6063: Spoken Language Processing. My name is 李思睿(lisirui) 223040027.

✨ Description

Add sgmse implementation

👨‍💻 Changes Proposed

  • add sgmse implementation

🧑‍🤝‍🧑 Who Can Review?

@HeCheng0625 @Adorable-Qin

✅ 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
Copy link
Collaborator

@Adorable-Qin , Hi Zihao, could you help review this pr about speech enhancement?

@HarryHe11
Copy link
Collaborator

You can see the egs/sgmse/README.md.This task is for the final project for AIR6063: Spoken Language Processing.My name is 李思睿(lisirui) 223040027.

Hi Sirui, Thank you so much for your helpful contribution! Could you provide us with some samples and also checkpoints that could showcase the effectiveness of your model?

@HarryHe11 HarryHe11 self-requested a review April 7, 2024 07:08
@lithr1
Copy link
Author

lithr1 commented Apr 7, 2024

You can see the egs/sgmse/README.md.This task is for the final project for AIR6063: Spoken Language Processing.My name is 李思睿(lisirui) 223040027.

Hi Sirui, Thank you so much for your helpful contribution! Could you provide us with some samples and also checkpoints that could showcase the effectiveness of your model?

My model's checkpoint has only been trained halfway compared to the model in the source file(it needs 400000steps), but I lack computing resources (the source file used eight cards). I only used one card and have been training 200000steps for five days, but it has been proven to be trainable. For detailed samples, please refer to the link
https://yiufjt4rn74.feishu.cn/docx/FOK8dfW9mo7AhyxJMDecQiFen7O?from=from_copylink

@yuantuo666
Copy link
Collaborator

You can see the egs/sgmse/README.md.This task is for the final project for AIR6063: Spoken Language Processing.My name is 李思睿(lisirui) 223040027.

Hi Sirui, Thank you so much for your helpful contribution! Could you provide us with some samples and also checkpoints that could showcase the effectiveness of your model?

My model's checkpoint has only been trained halfway compared to the model in the source file(it needs 400000steps), but I lack computing resources (the source file used eight cards). I only used one card and have been training 200000steps for five days, but it has been proven to be trainable. For detailed samples, please refer to the link https://yiufjt4rn74.feishu.cn/docx/FOK8dfW9mo7AhyxJMDecQiFen7O?from=from_copylink

Hi Sirui, the access right to the Feishu docs is not configured appropriately, could you provide one link with public access rights so we can check the details on the samples?
image

@yuantuo666
Copy link
Collaborator

To help us manage the PRs, I have attached a checklist on the first message. Feel free to add more specific information, like examples and changes to help us understand your contribution. Thanks!

@lithr1
Copy link
Author

lithr1 commented Apr 7, 2024

You can see the egs/sgmse/README.md.This task is for the final project for AIR6063: Spoken Language Processing.My name is 李思睿(lisirui) 223040027.

Hi Sirui, Thank you so much for your helpful contribution! Could you provide us with some samples and also checkpoints that could showcase the effectiveness of your model?

My model's checkpoint has only been trained halfway compared to the model in the source file(it needs 400000steps), but I lack computing resources (the source file used eight cards). I only used one card and have been training 200000steps for five days, but it has been proven to be trainable. For detailed samples, please refer to the link https://yiufjt4rn74.feishu.cn/docx/FOK8dfW9mo7AhyxJMDecQiFen7O?from=from_copylink

Hi Sirui, the access right to the Feishu docs is not configured appropriately, could you provide one link with public access rights so we can check the details on the samples? image

I have opened the access right,you can open the link again.

@lithr1 lithr1 requested a review from yuantuo666 April 8, 2024 04:49
@Adorable-Qin
Copy link
Collaborator

Hi @lithr1 !

Thank you for your efforts to improve Amphion.

However, the samples you attached do not sound as good as expected from the paper you are trying to reproduce. As what you said that your model may lack training or you need to scale up the dataset used during training, I recommend not to merge this PR until you can get a reasonable result. Then we could consider merging.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

5 participants