A Hybrid Speech Enhancement Algorithm for Voice Assistance Application

Round 1

Reviewer 1 Report

This paper proposed a hybridized speech enhancement algorithm to enhance the speech recognition accuracy. However, there are some problem to be addressed.

1) The abbreviation NSS needs to be explained the first time it appears.

2) The theoretical innovation is neither enough nor apparent.

3) An overview of the part 3 of the article appears in Part 2, paragraph 2, which can cause confusion.

4) Figure 4 is not illustrated and described.

5) There are symbol writing error in the 6th line of Algorithm 1.

6) The proposed model was only tested on the dataset that medical speech. What is the effect of applying this algorithm to other speech scenes?

7) There are many problems with the formatting, for example the first paragraph of part 4.2.1, table 2 and table 3.

8) The title of section 4.3 is performance analysis of HSEA, but why does the content become CHSEA?

9) The proposed algorithm is compared to typical methods, but it is not compared with the state of the art literature algorithms.

10) The number of references in this paper was limited and it may lead to an inadequate understanding of the current state of research.

Author Response

Dear Reviewer,

             Thank You for your valuable comments and recommendations for my manuscript. We have revised the manuscript as per your suggestions. 

Author Response File: Author Response.pdf

Reviewer 2 Report

The paper describes a speech recognition algorithm that combines the benefits of non-linear spectral subtraction and hidden markov model. The results indicate in reduction in word error rate and improved recognition accuracy. The experiments were conducted on medical transcription files. Overall the paper is organized well and the technical details are sufficiently explained. However, the reviewer recommends adding labels to the images explaining what each term means. For e.g. in Fig 3 EFV, REFV and NG should be explained either in the figure or its heading. Reviewer recommends providing more details on the dataset and may be a citation? What was the age group of the male and females. Was the audio recording done in a controlled environment or from a real medical scene? Additionally for the performance testing, how was the training and testing data split? Reviewer recommends adding these details so that it helps the readers understand the experimental design better.

Author Response

Dear Reviewer,

      Thank You for your valuable comments and recommendations. We have revised the manuscript according to your comments.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

the revised version is satisfactory, so i recommend it for acceptance.

please consider the state of the art literature algorithms for comparison.

Author Response

Dear Reviewer,

        Thank You for your valuable comments. The state of the art algorithms are compared and updated in a table format.  

Author Response File: Author Response.pdf