-
Notifications
You must be signed in to change notification settings - Fork 30
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
Proposal for Providing Ground Truth Data for Stuttering Speech Dataset #15
Comments
Hello,i'm pretty interested in your work , is it open sourced? How can i get access to it? |
Heyy i hope you are doing amazing , have you found the ground truth data for sep28k?? . If yes can you tell mw how you got it?. |
Hi @seblemaguer , @colincsl , I am currently engaged in stuttering-related speech research, and have been utilizing your dataset for our studies. During our utilization, we have observed that the stuttering dataset lacks accurate ground truth annotations. We understand that manually annotating ground truth for stuttering audio is a highly tedious and time-consuming task.
Therefore, we propose to offer our method to provide the necessary ground truth annotations for the dataset. Our technology primarily transforms stuttering audios into clear, stutter-free versions: starting by identifying stutter components within the audio and converting them into tokens, then using advanced large models for precise corrections, and finally reconstructing the audio with voice cloning technology to closely match the original recordings. In this process, repaired texts are also generated.
We believe these highly accurate texts could serve as the ground truth annotations for the dataset. We have processed the entire dataset and generated corresponding repaired texts. As an example, we have provided content from the beginning of the dataset segment “male-episode-4-with-joseph”.
The text was updated successfully, but these errors were encountered: