Black-Box Watermarking and Blockchain for IP Protection of Voiceprint Recognition Model

Round 1

Reviewer 1 Report

The ABSTRACT should let readers to instantly understand the methodology of the study and the benefits that this study can bring. 

The keywords " Copyright protection "  and " Model watermarking " appears too little.

Perhaps can have more new references in recent years. 

However, I suggest adding more references.

 

There should be a suggestion for current situation in the CONCLCUSION section.

 

Explain more why you want to get the results of these studies.

 

Please follow the format of the Journal.

The title of a paper should be clear and informative, and should reflect the aim and approach of the work. Your title is too long and should be reduced. A good title should not exceed 10 words.

Author Response

Please see the attachment

Author Response File: Author Response.pdf

Reviewer 2 Report

 

Reviewer(s)' Comments to Author:

This paper presents a black-box voiceprint recognition model protection framework that combines active and passive protection. It embeds key information into the Mel spectrogram to generate trigger samples that are difficult to detect and remove, and injects them into the host model as watermark W. Their approach enhances the protection performance of voiceprint recognition model copyrights. Experimental results show thatthe proposed method effectively protects voiceprint recognition model copyrights and restricts unauthorized access. 

In general, the idea of this paper technically makes sense; and the manuscript is easy to follow. Hence, I may suggest a minor revision for this work. However, the following issues should be addressed:

 

Detailed comments: 

1. The introduction should briefly indicate the main contributions of this paper in bullets, before the article structure at the end of the introduction section.

2.  Language in the whole manuscript requires improvements. There are some grammatical issues throughout the paper.  Some grammatical errors and expressions need to be further improved.

3. The proposed model should be compared with the known models by time

4. It is preferable to write the network configuration in a table. How did the authors set the parameters for the experimental results?

 

5. The paper does not discuss the limitations of the proposed approach, and the potential future directions for improvement.

 

6. High quality figures for Figure 2,3,4 are required.

7.  In Algorithm1, some subsript letters for the parameters M,X,and D are missing. kindly, revise it.

8. Page 8, Line 263, below Equation (5), there are some extra numbering ( 4. si and 5. wu)

9. Some sentences could be shortened.

10. Errors in using the singular and plural forms of nouns Errors in the singular and plural forms of nouns.

11. Readers would appreciate to see the authors’ implementation codes. Therefore, please upload the implementation codes to the public repositories, such as MATLAB File Exchange or GitHub, so that we can reproduce Tables and Figures of the paper

 

English language fine. Just, some little minor editing is required.

Author Response

Please see the attachment

Author Response File: Author Response.pdf