A Survey of Automatic Speech Recognition for Dysarthric Speech
Round 1
Reviewer 1 Report
The paper needs major re-organization, and I recommend Major revisions for this article. My details comments are attached here.
Comments for author File: 
Comments.pdf
The quality of the English language is satisfactory, at few places there are grammatical mistakes that needs to be addressed.
Author Response
Response to Reviewer 1 Comments
Point 1: Overall comments: The paper focuses on an important area of speech pathology and will be of great contribution to the readers. The paper needs some major re-organization of the sections to improve the readability of the users, and needs to include few diagrams showing evolution of ASR, methodology in trend or tables to summarize the trends in ASR. It will be great if authors can resubmit the paper after re-organizing and adding more relevant diagrams and tables.
Response 1: Thanks for your comments very much. Here we would response your comments point by point.
Abstract:
Point 2: This line is incorrect or imcomplete “Automatic speech recognition (ASR) is an effective way to help speakers with dysarthria communicate with other speakers.”
Response 2: Thanks for your comments. We modified this description. The details can be found in the second sentence highlighted in Section Abstract of the revised manuscript.
Point 3: Delete this line: We use “Dysarthric Speech 14 Recognition”, “Dysarthria Speech Recognition”, “Automatic Speech Recognition of Dysarthric 15 Speech” and “Automatic Speech Recognition of Dysarthria Speech” as key words to search papers 16 from Web of Science, Engineering Village and IEEE Xplore databases. Rather, right few of the main methods that contributed in the literature the most.
Response 3: Thanks for your suggestion. The line “We use … databases.” has been removed. Moreover, we supplemented some main methods of ASR for dysarthric speech.
Point 4: Not a relevant sentence in abstract: In this paper, according to the discussion of all authors, we select the representative papers and analyze the development of dysarthric speech recognition.
Response 4: Thanks for your suggestion. We have removed the line “In this paper, according to the discussion of all authors, we select the representative papers and analyze the development of dysarthric speech recognition.” in the abstract of revised manuscript.
Point 5: Please re-write abstract again and focus on the main algorithms/trends that you found in literature.
Response 5: Thanks for your suggestion. We have re-written the Abstract, the details can be found in highlighted parts of Abstract in revised manuscript.
Introduction:
Point 6: Dysarthia is a speech disorder not a language disorder. Please correct this in all of the article. E.g. Dysarthria as a language disorder only refers to neuromuscular disturbances concerning strength, speed, tone, steadiness or accuracy of the movements responsible for speech production.
Response 6: Thanks for your suggestion. This wrong description has been corrected as “speech disorder”
Point 7: Explain a bit about evolution of ASR and use some graphics to represent those trends.
Response 7: Response: Thanks for your suggestion. The evolution trend of ASR technologies has been supplemented in Section Introduction after the third paragraph in the revised manuscript.
Point 8: This section needs to be rewritten in a better organized way, e.g. you may subdivide the trends into before Deep learning and after deep learning, and compare the effects of each.
Response 8: Thanks for your suggestion. We have rewritten this section and subdivided the trends of ASR for dysarthric speech into “Before Deep Learning” and “During Deep Learning”. The details can be found in subsection 1.1 and 1.2 highlighted in Section Introduction.
Methods:
Point 9: There is no need of a separate section for methods. Reduce the content and fit the text into introduction, as this section has no discussion, and have no impact.
Response 9: Thanks for your suggestion. We reduced the original Section 2 and fit this part into Introduction. The details can be found in subsection 1.3 and 1.4.
Results:
Point 10: Rename this section something like : Summary of state-of-the-art or trends in ASR
Response 10: Thanks for your suggestion. The title of Results has been renamed to “Summary of the Evolutional Trends in ASR”.
Point 11: The first two paragraphs are not required, adds no knowledge to the reader and hence should be deleted.
Response 11: Thanks for your suggestion. The first two paragraphs have been removed.
Point 12: Subdivide this big section into few smaller sub sections to improve readability.
Response 12: Thanks for your suggestion. We divided the Section of databases into five subsections. The details can be found at the highlighted parts under Table 1.
Point 13: Example: a subsection for databases available and add some relevant discussion and tables in this subsection.
Response 13: Thanks for your suggestion. We added some relevant discussions and tables for describing databases in this subsection. The details can be found in the highlighted parts and subsection from 2.1.1 to 2.1.5.
Point 14: Add the different databases into bullet points or numbers rather than paragraphs, it improves the readability.
Response 14: Thanks for your suggestion. We divided different databases into several subsections. The details can be found from subsection 2.1.1 to 2.1.5.
Point 15: You may refer one of the articles below to get an idea about sections and images you may include to improve the quality of the review.
- Trends in audio signal feature extraction methods – ScienceDirect
- 21554.pdf (aes.org)
- Trends in biomedical signal feature extraction – ScienceDirect
Response 15: Thanks for your kind suggestion very much.
Author Response File: Author Response.pdf
Reviewer 2 Report
The manuscript misses several important parts to improve its comprehension and credibility.
During the abstract, it is better to avoid talking about used keywords and searched databases; instead, we focus on the importance of the ASR to help speakers with dysarthria and then present the lessons learned after completing the survey.
The authors briefly summarize existing works in three lines in the introduction; however, the manuscript must include a related work section describing similar surveys and how their survey outperforms them.
The name of Section 3 is not convenient. It must be representative, such as "ASR Approaches for Dysarthric Speech".
Section 3 is clear. In other words, it must contain mind-map diagrams to summarize each subsection. Thus, the reader could find the required information quickly.
Section 4 is not comprehensive; it must contain a comparative study with existing surveys, the lessons learned from their research, and how to improve the current ASR techniques.
Moderate editing of English language required
Author Response
Response to Reviewer 2 Comments
Point 1: The manuscript misses several important parts to improve its comprehension and credibility.
Response 1: Thanks for your comments very much. Here we would response your comments point by point.
Point 2: During the abstract, it is better to avoid talking about used keywords and searched databases; instead, we focus on the importance of the ASR to help speakers with dysarthria and then present the lessons learned after completing the survey.
Response 2: Thanks for your suggestion. We re-write the Section Abstract to make it be more readable.
Point 3: The authors briefly summarize existing works in three lines in the introduction; however, the manuscript must include a related work section describing similar surveys and how their survey outperforms them.
Response 3: Thanks for your suggestion. We added the previous reviews related to our topics and supplemented the description that our survey outperforms the previous survey. The details can be found in the highlighted parts from line 107 to line 116.
Point 4: The name of Section 3 is not convenient. It must be representative, such as "ASR Approaches for Dysarthric Speech".
Response 4: Thanks for your suggestion. We renamed the title of Section 3 into “Summary of the Evolutional Trends in ASR”.
Point 5: Section 3 is clear. In other words, it must contain mind-map diagrams to summarize each subsection. Thus, the reader could find the required information quickly.
Response 5: Thanks for your suggestion. We supplemented some mind-map diagrams to summarize each subsection. For example, Figure 3~6 were added in the revised manuscript.
Point 6: Section 4 is not comprehensive; it must contain a comparative study with existing surveys, the lessons learned from their research, and how to improve the current ASR techniques.
Response 6: Thanks for your suggestion. We re-write this section to improve the readability of this part. After that, the comparative study with existing surveys, the lessons learned from their researches, and how to improve the current ASR techniques are all contained.
Author Response File: Author Response.pdf
Round 2
Reviewer 1 Report
Authors have incorporated all the suggestions provided in R1. A great amount of improvement in the revised manuscript now. I can comfortably recommend this version for publication.
The English is satisfactory, just need to be proofread for grammatical errors.
Reviewer 2 Report
The manuscript has been improved. All my comments were adressed.
Minor English editing.
