Accelerated Conjugate Gradient for Second-Order Blind Signal Separation

Round 1

Reviewer 1 Report

Please see the attachment.

Comments for author File: Comments.pdf

Author Response

We would like to thank Reviewer for the insightful comments on the paper. We highly appreciate the significant time and knowledge invested. We have revised the paper according to the suggested comments.

Thank you

Kind Regards,

Hai Huyen Dam and Sven Nordholm

Author Response File: Author Response.pdf

Reviewer 2 Report

Contributions:

This paper proposes an adaptive algorithm for the second-order blind signal separation problem by an accelerated gradient and a conjugate gradient method. My comments are given below:

1. The major contribution is given in (7) to (11). They are heuristic. The novelty is limited.
2. (Page 2) The operator “offdiag” should be defined.
3. (Page 2) In (3), I think E should be R_y.
4. (Line 63 on page 2)I think r(w) is only a normalized factor rather than a weighting function.
5. (Page 3) Equation (4) is not well presented. The left curly bracket should be removed.
6. (Page 3) The equation number (6) should be removed.
7. (Page 3) Equation (5) is not correct. It should be gradient f, rather than gradient W.
8. (Page 3) How does the operator off work in (5)?
9. (Page 3) In (7), k is the iteration index. However, it is a time index in (1). The usage of the parameter k is inconsistent.
10. (Page 3) The delta^hat is not defined.
11. (Page 3) In (9), the equation is independent of i.
12. (Page 4) the parameter tao in (12) should be defined.
13. (Page 4) “k+1_>k” should be revised as “k=k+1”.
14. (Page 5)Equation numbers should be given for the SIR and SNR. 
15. The titles in Figs. 2 to 5 should be removed.
16. The sub-grid lines in Figs. 2-5 and tables 1-4 should be removed.
17. English written should be further improved.

A.(Line 13 on page 1) “)” should be removed.

B.(Line 2 after section 2) “at L microphones …” should be revised as “with L microphones …”.

C.(Line 4 before (2)) “Denote by W…” should be revised as “Denoted by W…”.

D.(Line 128 on page 5)”Figs. 2 and 3 shows…” should be revised as “Figurs 2 and 3 shows…”.

Author Response

We would like to thank Reviewer for the insightful comments on the paper. We highly appreciate the significant time and knowledge invested. We have revised the paper according to the suggested comments.

Thank you

Kind Regards,

Hai Huyen Dam and Sven Nordholm

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

The authors have responded to the reviewer's remarks. But there are some issues that I think the author doesn't explain clearly.

1.      Comment 6: The SIR can be calculated by considering one speech as the source and the other as the interference. Then how to calculate the SNR? How to separate the interference from the noise?

2.      Comment 8: The authors do not explain how the difference between image and speech is handled when applying the proposed algorithm.

3.      Comment 11: According to Tables 3 and 4, doesn't the output of the accelerated conjugate gradient method have higher SIR and SNR?

Author Response

Dear reviewer

We would like to thank the reviewer for the insightful comments on the paper. We highly appreciate the significant time and knowledge invested. We have revised the paper according to the suggested comments.

Thank you

Author Response File: Author Response.pdf

Reviewer 2 Report

1. The authors have improved the quality of this paper. My comments are given below:

   1. In my previous comment 1: “The major contribution is given in (7) to (11). They are heuristic. The novelty is limited.”. The authors did not reply to my comment.

   2. In my previous comment 8: “How does the operator off work in (5)?”. The authors did not exactly reply to the “off” operator's operation.

Author Response

Dear Reviewer

We would like to thank the reviewer for the insightful comments on the paper. We highly appreciate the significant time and knowledge invested. We have revised the
paper according to the suggested comments.

Thank you

Kind Regards,

Hai Huyen Dam

Author Response File: Author Response.pdf

Round 3

Reviewer 1 Report

The author have improved the paper and I suggest the paper for publication in the current journal. But I would like the author to answer my questions.

Comment 2：” The main difference between image and speech problems lies in the cost-function.” How do the two cost functions reflect the difference between image and speech in time and spatial latitude?

Reviewer 2 Report

The authors have improved the quality of this paper. I suggest accepting this paper for publication.