Next Article in Journal
Mitigating Thermal Side-Channel Vulnerabilities in FPGA-Based SiP Systems Through Advanced Thermal Management and Security Integration Using Thermal Digital Twin (TDT) Technology
Previous Article in Journal
SS-Trans (Single-Stream Transformer for Multimodal Sentiment Analysis and Emotion Recognition): The Emotion Whisperer—A Single-Stream Transformer for Multimodal Sentiment Analysis
Previous Article in Special Issue
Cross-Domain Document Summarization Model via Two-Stage Curriculum Learning
 
 
Article
Peer-Review Record

Improving Top-Down Attention Network in Speech Separation by Employing Hand-Crafted Filterbank and Parameter-Sharing Transformer

Electronics 2024, 13(21), 4174; https://doi.org/10.3390/electronics13214174
by Aye Nyein Aung and Jeih-weih Hung *
Reviewer 1: Anonymous
Reviewer 2: Anonymous
Reviewer 3: Anonymous
Reviewer 4: Anonymous
Electronics 2024, 13(21), 4174; https://doi.org/10.3390/electronics13214174
Submission received: 15 September 2024 / Revised: 17 October 2024 / Accepted: 23 October 2024 / Published: 24 October 2024
(This article belongs to the Special Issue Natural Language Processing Method: Deep Learning and Deep Semantics)

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

electronics-3235522-peer-review-v1

 

The manuscript presents very interesting research results. The authors correctly presented the conducted research and the obtained results.

 

The manuscript requires only a slight expansion of the presented research results in chapter "5.4. Experimental Results and Discussion":

- this chapter should be expanded and more examples of research results should be presented together with their analysis,

- the resolution of the research results in Figure 6 is too poor

- the research results in Figure 6 should be characterized in more scientific detail

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

Refer to attached file for comments.

Comments for author File: Comments.pdf

Comments on the Quality of English Language

ok

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

I would like to draw your attention to the article titled "Improving Top-Down Attention Network in Speech Separation by Employing Hand-Crafted Filterbank and Parameter-Sharing Transformer."

Firstly, I would like to commend you on the thorough and systematic preparation of the text, as its organization and clarity are commendable.

I have outlined several suggestions for enhancing the text:

1. It would be beneficial to enrich the first and second paragraphs of the introduction, specifically on lines 29-31 and 40-44, with additional references. A more detailed exploration of the neuroanatomical processes involved in the top-down effect related to attention and speech separation would significantly strengthen your argument and provide a more robust foundation for your claims. Additionally, reinforcing the bibliography at the conclusion of the second paragraph would be appropriate.

2. When introducing abbreviations, such as the one on line 46, it is advisable to provide their full forms to facilitate reader comprehension. The introduction should also encompass existing knowledge and the overall relationship between the attentional network and auditory processing.

3. It is crucial to address the absence of bibliographic references in certain sections of the text, particularly in section 4.2, by including the necessary citations.

4. Your presentation of the results and the model is well-articulated.

5. In your conclusions, it is important to clearly delineate the novel contributions of your study, the modifications made to existing models, and to acknowledge any limitations that may exist.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 4 Report

Comments and Suggestions for Authors

This paper addresses the challenge of separating overlapping speech signals by improving the separation capabilities of the Top-Down Attention Network (TDANet) method while ensuring a favorable balance between performance and resource efficiency. Two modifications for TDANet are proposed and combined.

The manuscript is well written and technically sound. The theory seems correct and the experimental results and comparisons are good.

However, the authors need to better clarify the combination of the two proposed methods and, overall, the novelty of the presented study.

Finally, authors should check their manuscript for errors and typos, e.g.  "TDANet with MF-GTF, reveals a 0.34 absolute improvement in SI-SDRi compared to the original TDANet ..." at page 10, line 320: is the absolute improvement 0.37, or did I miss something?

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 2 Report

Comments and Suggestions for Authors

No more comments. Ready for acceptance and publication.

Reviewer 4 Report

Comments and Suggestions for Authors

The authors have addressed enough of my comments and suggestions.

In addition, the authors further improved this manuscript by answering reviewers’ questions and revising the manuscript.

Consequently, this paper now can be accepted in its present form.

Back to TopTop