search

facebookresearch / audioseal

Localized watermarking for AI-generated speech audios, with SOTA on robustness and very fast detector
MIT License
466 stars 56 forks source link

Unable to identify previously audioseal watermarked files #19

Closed KabaTubare closed 9 months ago

KabaTubare commented 9 months ago

Here is the notebook code: https://colab.research.google.com/gist/KabaTubare/13d0ef6ecb132a8c1a73bb143352c10c/audioseal-inference.ipynb

KabaTubare commented 9 months ago
Screenshot 2024-02-29 at 8 27 29

As you can see, this was a audioseal watermarked file that was falsely id'd as not having a watermark. Note this function of detection has been de-coupled from your original code.

pierrefdz commented 9 months ago

Hi @KabaTubare,

We appreciate your enthusiasm for improving the project, but it's important to remember that the issue tracker is intended for reporting specific bugs or feature requests. Would you mind reading the paper first? This will answer most of your questions.

Thank you for your understanding and cooperation.

antoine-tran commented 9 months ago

Hi @KabaTubare , is your concern with the detector resolved ? If so, I would like to close this issue

KabaTubare commented 9 months ago

Hey there!

Yes, I have a demo up and running on HF already🙌🏽🖖🏾

Kind regards,

Troy Woodson

On Sat 2. Mar 2024 at 2.19 PM, Tuan Tran @.***> wrote:

Hi @KabaTubare https://github.com/KabaTubare , is your concern with the detector resolved ? If so, I would like to close this issue

— Reply to this email directly, view it on GitHub https://github.com/facebookresearch/audioseal/issues/19#issuecomment-1974782351, or unsubscribe https://github.com/notifications/unsubscribe-auth/A7AVNVVEJUFET6G2U7OIRXDYWG7WRAVCNFSM6AAAAABD7KZNP2VHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTSNZUG44DEMZVGE . You are receiving this because you were mentioned.Message ID: @.***>