CC
Subject
IEEE Access - Decision on Manuscript ID Access-2019-59962
Body
31-Dec-2019
Dear Dr. Li:
I am writing to you in regards to manuscript # Access-2019-59962 entitled "A Hybrid Ensemble-Learning method for Music Classification and Similarity Measurement" which you submitted to IEEE Access.
In view of the criticisms of the reviewer(s) found at the bottom of this letter, your manuscript has not been recommended for publication in IEEE Access.
We do encourage you to revise and resubmit your article once you have addressed the concerns and criticisms of the reviewers. I believe they have added good insight on how to further improve your article. IEEE Access has a binary peer review process that does not allow revisions. Therefore, in order to uphold quality to IEEE standards, an article is rejected even if it requires minor edits.
Please revise your manuscript based on reviewers’ feedback and resubmit; elaborate on your points and clarify with references, examples, data, etc. If you do not agree with the reviewers’ views, then include your arguments in the updated manuscript. Also, note that if a reviewer suggested references, you should only add ones that will make your article better and more complete. Recommending references to specific publications is not appropriate for reviewers and you should report excessive cases to ieeeaccessEIC@ieee.org.
We highly recommend that you review the grammar one more time before resubmitting. IEEE offers a 3rd party service for language polishing, which you may utilize for a fee: https://www.aje.com/c/ieee (use the URL to claim a 10% discount).
Please be advised that authors are only permitted to resubmit their article ONCE. If the updated manuscript is determined not to have addressed all of the previous reviewers’ concerns, the article may be rejected and no further resubmissions will be allowed.
When resubmitting, please submit as a new manuscript and include the following 3 files:
1) A document containing your response to reviewers from the previous peer review. The “response to reviewers” document (template attached) should have the following regarding each comment: a) Reviewer’s concern, b) your response to the concern, c) your action to remedy the concern. The document should be uploaded with your manuscript files as a “Supplemental File for Review.”
2) Your updated manuscript with all your individual changes highlighted, including grammatical changes (e.g. preferably with the yellow highlight tool within the pdf file). This file should be uploaded with your manuscript files as a “Supplemental File for Review”.
3) A clean copy of the final manuscript (without highlighted changes) should be submitted as the “Formatted (Double Column) Main File – PDF Document Only.”
Finally, in your cover letter, please indicate if you would like us to assign your article to the same or different reviewers and we will do our best to accommodate your request.
We sincerely hope you will update your manuscript and resubmit soon. Please contact me if you have any questions.
Additional Questions:
Does the paper contribute to the body of knowledge?: No!
The authors have publised a similar article with title on 2016 IEEE International Conference on e-Business Engineering.
So I think this paper have nothong newer contribution to the body of knowledge.
Is the paper technically sound?: Yes.
The paper belongs to the field of music information retrieval combined with deep learning applications.
Is the subject matter presented in a comprehensive manner?: Yes.
Paper writing is good.
Are the references provided applicable and sufficient?: Yes.
The references are sufficient and reasonable.
Reviewer: 2
Recommendation: Reject (do not encourage resubmit)
Comments:
This paper comes across as the first author's phd work condensed into a single paper. whether that is correct or not, there is far too much going on here and it is all poorly described.
If a reader is to be able to believe your results they have to understand what you have done to get them. For example, there is no evidence that the MDL is a complete representation of music - that in itself could form one paper. The same is true for MML. when it comes to the description of Machine learning etc, it's not clear why the specific aproaches have been chosen, or if better choices could have been made.
I suggest you try out one story from this complicated work and write a paper on that only, rather than telling too many stories inadequately.
Additional Questions:
Does the paper contribute to the body of knowledge?: No.
Is the paper technically sound?: It is not straightforward to determine if it is technically sound because it is very poorly written. Because the paper's description of related work is so poorly done and sparse, and there is no description whatever of musical theory, it is not easy or even moderately difficult to decide whether or not the definitions of the MDL and MML are valid for music description.
In my view, there is some similarity to aspect of the SAOL language from the 90s, which was part of MPEG4. But this is basically a guess because the descriptions of MDL and MML are so obscure as to be impenetrable. I would suggest that the authors also look at the works of Wiggins et al, specifically the CHARM music description.
Is the subject matter presented in a comprehensive manner?: Most definitely not. it is hard to understand throughout. There are too many sentences to list here that seem to make no sense.
Are the references provided applicable and sufficient?: No. There is a lot of work cited from the 90s and 2000s, but not much from recent works, and almost nothing from the ISMIR conferences which are essentially the major venue for the type of work that this purports to be.
Reviewer: 3
Recommendation: Reject (do not encourage resubmit)
Comments:
The paper brings an idea which may sound attractive in the first moment. The authors claim that it is the first time that anyone approaches similarity-based music retrieval using the scores and the audio in a single model. That is likely to be correct since the idea is not good. I can see how someone may need to know which composition represents a score sheet (or a MIDI file) or an unknown audio playing right now (like Shazam does). However, I believe that having both score and audio in hands, at the same time that the user does not know which song is that, is not usual at all.
Besides, the writing and presentation are bad. I suggest, case the authors intend to submit the manuscript again or somewhere else, to proofread the text and significantly improve the quality of the figures.
Additional Questions:
Does the paper contribute to the body of knowledge?: No. I explain why in the general comments.
Is the paper technically sound?: Yes.
Is the subject matter presented in a comprehensive manner?: Not much. There is plenty of issues with the writing and the presentation.
Are the references provided applicable and sufficient?: Yes.
Reviewer: 4
Recommendation: Reject (update and resubmit encouraged)
Comments:
In this paper, a hybrid method using two music representation(Music Definition Language (MDL), Music Manipulation Language (MML)) and ensemble-learning for music classification
and similarity measurement is proposed. This is an extension of authors' previous work 'Towards a Hybrid Deep-Learning Method for Music Classification and Similarity Measurement', which already discussed the MDL and MML for music classification
and similarity measurement. So the novel part of this work is reinforcement based ensemble learning.
1.In this paper, all relevant methods, algorithms are listed, but I can't get the reason or intuition to combine reinforcement based ensemble learning with MDL and MML.
2.The wrtting should be further improved, should be more logically and emphasize the novel parts.
3.Most of figures are in low-resolution and hard to understand.
4.Descriptions of algorithm should be formatted.
5.In expriments, results of other methods are preferred for comparision.
Additional Questions:
Does the paper contribute to the body of knowledge?: Partially
Is the paper technically sound?: Partially
Is the subject matter presented in a comprehensive manner?: Yes
Are the references provided applicable and sufficient?: Yes
Reviewer: 5
Recommendation: Reject (update and resubmit encouraged)
Comments:
It would be better to highlight the difference between the paper and the earlier work in [35]
It is critical to describe how to convert music audio into the A-MDL data model, as well as how it performs in experiments. The data size in the experiment should be high enough, say in thousands of songs.
Page 7 of 17, line 14: M-MDL should be A-MDL ?
Page 7 of 17, line 35: T-MDL should be T-MML?
Additional Questions:
Does the paper contribute to the body of knowledge?: Yes, generally the paper contributes to the body of knowledge.
Is the paper technically sound?: Not quite so.
The paper claimed the developed system can handle both symbolic music and music audio. The proposed data model captures the musical content of the music (S-MDL for symbolic and A-DML for audio). The music classification/search/identification is carried out based on this data model. Unfortunately, the important component of converting a music audio recording into the proposed data model (A-MDL w/ T-MML) is missing in the paper. The performance of such converting is critical for the performance of the system. This component is supposed to do acoustic signal analysis and musical feature extraction.
In the experiment (V.B.), it is not clear how the A-DML is obtained from the 10 songs (either original or variations). On top of that, the size of data in the experiment is too small for the task of music plagiarism detection.
Is the subject matter presented in a comprehensive manner?: Music plagiarism identification based on audio is an important application. It is also technically challenging. The extraction feature, such as the proposed data model, for music audio should be compact and adequately capture the musical content and also resilient to noise and variations.
The mentioned earlier audio fingerprint method is meant for identification of a particular music recording with distortion in terms of environmental noise in the query snippet. Music variation would generate way too different fingerprint from the original audio for the method to work.
Melody-based music audio search is more relevant to the task. However, melody extraction from music audio (or converting music audio to A-MDL) is not addressed in the paper.
Are the references provided applicable and sufficient?: Yes.
Reviewer: 6
Recommendation: Reject (update and resubmit encouraged)
Comments:
(1) Just like you have given the definitions of the 6 tuples of S-MDL in bullet form it will be better to define the 3 tuples of the A-MDL in a similar way before defining the A-MDL itself.
(2) You have stated that in section IV subsection C that "input melody to MUCASM is encoded with the proposed scheme MDL" . It should be made clear whether it is done manually or using some automated algorithm or tool. It appears that you have done it manually because no mention of algorithm is there. If so it should be outlined. And the musical skills of the person doing it manually should be verified and outline in the manuscript.
(2) If the manual conversion of a piece of music to the so defined MDL and MML has been done then the title of the paper should be little modified to either
(a) A Hybrid Ensemble-Learning Method for Music DATA Classification and Similarity. OR (b) A Hybrid Ensemble-Learning Method for Music REPRESENTATION Classification and Similarity.
Because the present title reflects the presence of an acoustic waveform file to the input of the system which is not the case in my view.
Please clarify it. The Machine learning part is nicely presented.
Additional Questions:
Does the paper contribute to the body of knowledge?: Yes, New Music Representations have been proposed.
Is the paper technically sound?: Yes, but the title is little misleading.
Is the subject matter presented in a comprehensive manner?: Yes quite comprehensive presentation. Only some clarifications have to be included regarding the method of conversion of acoustic waveform to MDL and MML representation.
Are the references provided applicable and sufficient?: Yes.
Reviewer: 7
Recommendation: Accept (minor edits)
Comments:
The paper is clearly written and the approach is also novelty. It can be accepted after minor revision.
Additional Questions:
Does the paper contribute to the body of knowledge?: Yes, make good contributions to the body of knowledge.
Is the paper technically sound?: Yes, the paper is technically sound and is of very high quality.
Is the subject matter presented in a comprehensive manner?: Very clearly.
Are the references provided applicable and sufficient?: Yes.
If you have any questions, please contact article administrator: Mr. Nishant Shukla n.shukla@ieee.org
Attached Files
* IEEE-Access-Response-to-Reviewers-template-1.28.19.doc，
我按照网站模板修改了，但投稿一直回复我：Thank you for your submission of Manuscript ID Access-2020-32471 entitled "……” which you submitted to IEEE Access. IEEE Access requires that manuscripts are submitted as a PDF in a double column, single-spaced format using a mandatory IEEE Access template. Additionally, submissions should have the traditional IEEE format of an abstract, introduction, full explanation of background and history, a detailed explanation of recent findings, a conclusion, and a list of references. We have returned the article to your Author Center as a draft so that you can make the necessary corrections, and then resubmit. You may find additional information regarding submission and formatting requirements for IEEE Access at https://ieeeaccess.ieee.org/submitting-an-article/ Please also use this opportunity to carefully review and update the grammar before resubmitting since proper grammar is a requirement for publication in IEEE Access. When you are ready to resubmit, go to https://mc.manuscriptcentral.com/ieee-access and log in to your Author Center. Click on "Unsubmitted and Manuscripts in Draft," and then click on "Continue" located next to the manuscript number. Please be sure to make the necessary updates before following the steps to resubmit your manuscript. We look forward to your resubmission. Sincerely, IEEE Access Editorial Office.
求问怎么办
返回小木虫查看更多
京公网安备 11010802022153号
access不是有模板吗，包括word和latex两种，肯定要按照模板排版
自己投过一次，没有中，两个审稿人，一个accept一个reject
刚看了一下自己审到的一篇论文：6个审稿人。。。
Sent
31-Dec-2019
From
To
CC
Subject
IEEE Access - Decision on Manuscript ID Access-2019-59962
Body
31-Dec-2019
Dear Dr. Li:
I am writing to you in regards to manuscript # Access-2019-59962 entitled "A Hybrid Ensemble-Learning method for Music Classification and Similarity Measurement" which you submitted to IEEE Access.
In view of the criticisms of the reviewer(s) found at the bottom of this letter, your manuscript has not been recommended for publication in IEEE Access.
We do encourage you to revise and resubmit your article once you have addressed the concerns and criticisms of the reviewers. I believe they have added good insight on how to further improve your article. IEEE Access has a binary peer review process that does not allow revisions. Therefore, in order to uphold quality to IEEE standards, an article is rejected even if it requires minor edits.
Please revise your manuscript based on reviewers’ feedback and resubmit; elaborate on your points and clarify with references, examples, data, etc. If you do not agree with the reviewers’ views, then include your arguments in the updated manuscript. Also, note that if a reviewer suggested references, you should only add ones that will make your article better and more complete. Recommending references to specific publications is not appropriate for reviewers and you should report excessive cases to ieeeaccessEIC@ieee.org.
We highly recommend that you review the grammar one more time before resubmitting. IEEE offers a 3rd party service for language polishing, which you may utilize for a fee: https://www.aje.com/c/ieee (use the URL to claim a 10% discount).
Please be advised that authors are only permitted to resubmit their article ONCE. If the updated manuscript is determined not to have addressed all of the previous reviewers’ concerns, the article may be rejected and no further resubmissions will be allowed.
When resubmitting, please submit as a new manuscript and include the following 3 files:
1) A document containing your response to reviewers from the previous peer review. The “response to reviewers” document (template attached) should have the following regarding each comment: a) Reviewer’s concern, b) your response to the concern, c) your action to remedy the concern. The document should be uploaded with your manuscript files as a “Supplemental File for Review.”
2) Your updated manuscript with all your individual changes highlighted, including grammatical changes (e.g. preferably with the yellow highlight tool within the pdf file). This file should be uploaded with your manuscript files as a “Supplemental File for Review”.
3) A clean copy of the final manuscript (without highlighted changes) should be submitted as the “Formatted (Double Column) Main File – PDF Document Only.”
Finally, in your cover letter, please indicate if you would like us to assign your article to the same or different reviewers and we will do our best to accommodate your request.
We sincerely hope you will update your manuscript and resubmit soon. Please contact me if you have any questions.
Thank you for your interest in IEEE Access.
Sincerely,
Dr. HALUK EREN
Associate Editor, IEEE Access
he.edu.tr@gmail.com, heren@firat.edu.tr
Reviewer(s)' Comments to Author:
Reviewer: 1
Recommendation: Reject (do not encourage resubmit)
Comments:
I think this just a copy of your work of 2016.
Please tell the difference of the two paper.
TITLE: A Hybrid Ensemble-Learning method for Music Classification and Similarity Measurement
AUTHORS: Li, Hanchao; Fei, Xiang; Chao, Kuo-Min; Yang, Ming; He, Chaobo
Towards A Hybrid Deep-Learning method for Music Classification and Similarity Measurement
AUTHORS: Li, Hanchao; Fei, Xiang; Chao, Kuo-Min; Yang, Ming; He, Chaobo
Additional Questions:
Does the paper contribute to the body of knowledge?: No!
The authors have publised a similar article with title on 2016 IEEE International Conference on e-Business Engineering.
So I think this paper have nothong newer contribution to the body of knowledge.
Is the paper technically sound?: Yes.
The paper belongs to the field of music information retrieval combined with deep learning applications.
Is the subject matter presented in a comprehensive manner?: Yes.
Paper writing is good.
Are the references provided applicable and sufficient?: Yes.
The references are sufficient and reasonable.
Reviewer: 2
Recommendation: Reject (do not encourage resubmit)
Comments:
This paper comes across as the first author's phd work condensed into a single paper. whether that is correct or not, there is far too much going on here and it is all poorly described.
If a reader is to be able to believe your results they have to understand what you have done to get them. For example, there is no evidence that the MDL is a complete representation of music - that in itself could form one paper. The same is true for MML. when it comes to the description of Machine learning etc, it's not clear why the specific aproaches have been chosen, or if better choices could have been made.
I suggest you try out one story from this complicated work and write a paper on that only, rather than telling too many stories inadequately.
Additional Questions:
Does the paper contribute to the body of knowledge?: No.
Is the paper technically sound?: It is not straightforward to determine if it is technically sound because it is very poorly written. Because the paper's description of related work is so poorly done and sparse, and there is no description whatever of musical theory, it is not easy or even moderately difficult to decide whether or not the definitions of the MDL and MML are valid for music description.
In my view, there is some similarity to aspect of the SAOL language from the 90s, which was part of MPEG4. But this is basically a guess because the descriptions of MDL and MML are so obscure as to be impenetrable. I would suggest that the authors also look at the works of Wiggins et al, specifically the CHARM music description.
Is the subject matter presented in a comprehensive manner?: Most definitely not. it is hard to understand throughout. There are too many sentences to list here that seem to make no sense.
Are the references provided applicable and sufficient?: No. There is a lot of work cited from the 90s and 2000s, but not much from recent works, and almost nothing from the ISMIR conferences which are essentially the major venue for the type of work that this purports to be.
Reviewer: 3
Recommendation: Reject (do not encourage resubmit)
Comments:
The paper brings an idea which may sound attractive in the first moment. The authors claim that it is the first time that anyone approaches similarity-based music retrieval using the scores and the audio in a single model. That is likely to be correct since the idea is not good. I can see how someone may need to know which composition represents a score sheet (or a MIDI file) or an unknown audio playing right now (like Shazam does). However, I believe that having both score and audio in hands, at the same time that the user does not know which song is that, is not usual at all.
Besides, the writing and presentation are bad. I suggest, case the authors intend to submit the manuscript again or somewhere else, to proofread the text and significantly improve the quality of the figures.
Additional Questions:
Does the paper contribute to the body of knowledge?: No. I explain why in the general comments.
Is the paper technically sound?: Yes.
Is the subject matter presented in a comprehensive manner?: Not much. There is plenty of issues with the writing and the presentation.
Are the references provided applicable and sufficient?: Yes.
Reviewer: 4
Recommendation: Reject (update and resubmit encouraged)
Comments:
In this paper, a hybrid method using two music representation(Music Definition Language (MDL), Music Manipulation Language (MML)) and ensemble-learning for music classification
and similarity measurement is proposed. This is an extension of authors' previous work 'Towards a Hybrid Deep-Learning Method for Music Classification and Similarity Measurement', which already discussed the MDL and MML for music classification
and similarity measurement. So the novel part of this work is reinforcement based ensemble learning.
1.In this paper, all relevant methods, algorithms are listed, but I can't get the reason or intuition to combine reinforcement based ensemble learning with MDL and MML.
2.The wrtting should be further improved, should be more logically and emphasize the novel parts.
3.Most of figures are in low-resolution and hard to understand.
4.Descriptions of algorithm should be formatted.
5.In expriments, results of other methods are preferred for comparision.
Additional Questions:
Does the paper contribute to the body of knowledge?: Partially
Is the paper technically sound?: Partially
Is the subject matter presented in a comprehensive manner?: Yes
Are the references provided applicable and sufficient?: Yes
Reviewer: 5
Recommendation: Reject (update and resubmit encouraged)
Comments:
It would be better to highlight the difference between the paper and the earlier work in [35]
It is critical to describe how to convert music audio into the A-MDL data model, as well as how it performs in experiments. The data size in the experiment should be high enough, say in thousands of songs.
Page 7 of 17, line 14: M-MDL should be A-MDL ?
Page 7 of 17, line 35: T-MDL should be T-MML?
Additional Questions:
Does the paper contribute to the body of knowledge?: Yes, generally the paper contributes to the body of knowledge.
Is the paper technically sound?: Not quite so.
The paper claimed the developed system can handle both symbolic music and music audio. The proposed data model captures the musical content of the music (S-MDL for symbolic and A-DML for audio). The music classification/search/identification is carried out based on this data model. Unfortunately, the important component of converting a music audio recording into the proposed data model (A-MDL w/ T-MML) is missing in the paper. The performance of such converting is critical for the performance of the system. This component is supposed to do acoustic signal analysis and musical feature extraction.
In the experiment (V.B.), it is not clear how the A-DML is obtained from the 10 songs (either original or variations). On top of that, the size of data in the experiment is too small for the task of music plagiarism detection.
Is the subject matter presented in a comprehensive manner?: Music plagiarism identification based on audio is an important application. It is also technically challenging. The extraction feature, such as the proposed data model, for music audio should be compact and adequately capture the musical content and also resilient to noise and variations.
The mentioned earlier audio fingerprint method is meant for identification of a particular music recording with distortion in terms of environmental noise in the query snippet. Music variation would generate way too different fingerprint from the original audio for the method to work.
Melody-based music audio search is more relevant to the task. However, melody extraction from music audio (or converting music audio to A-MDL) is not addressed in the paper.
Are the references provided applicable and sufficient?: Yes.
Reviewer: 6
Recommendation: Reject (update and resubmit encouraged)
Comments:
(1) Just like you have given the definitions of the 6 tuples of S-MDL in bullet form it will be better to define the 3 tuples of the A-MDL in a similar way before defining the A-MDL itself.
(2) You have stated that in section IV subsection C that "input melody to MUCASM is encoded with the proposed scheme MDL" . It should be made clear whether it is done manually or using some automated algorithm or tool. It appears that you have done it manually because no mention of algorithm is there. If so it should be outlined. And the musical skills of the person doing it manually should be verified and outline in the manuscript.
(2) If the manual conversion of a piece of music to the so defined MDL and MML has been done then the title of the paper should be little modified to either
(a) A Hybrid Ensemble-Learning Method for Music DATA Classification and Similarity. OR (b) A Hybrid Ensemble-Learning Method for Music REPRESENTATION Classification and Similarity.
Because the present title reflects the presence of an acoustic waveform file to the input of the system which is not the case in my view.
Please clarify it. The Machine learning part is nicely presented.
Additional Questions:
Does the paper contribute to the body of knowledge?: Yes, New Music Representations have been proposed.
Is the paper technically sound?: Yes, but the title is little misleading.
Is the subject matter presented in a comprehensive manner?: Yes quite comprehensive presentation. Only some clarifications have to be included regarding the method of conversion of acoustic waveform to MDL and MML representation.
Are the references provided applicable and sufficient?: Yes.
Reviewer: 7
Recommendation: Accept (minor edits)
Comments:
The paper is clearly written and the approach is also novelty. It can be accepted after minor revision.
Additional Questions:
Does the paper contribute to the body of knowledge?: Yes, make good contributions to the body of knowledge.
Is the paper technically sound?: Yes, the paper is technically sound and is of very high quality.
Is the subject matter presented in a comprehensive manner?: Very clearly.
Are the references provided applicable and sufficient?: Yes.
If you have any questions, please contact article administrator: Mr. Nishant Shukla n.shukla@ieee.org
Attached Files
* IEEE-Access-Response-to-Reviewers-template-1.28.19.doc，
好像所有的期刊都有模板吧，你之前投的论文都没有模板吗？
我按照网站模板修改了，但投稿一直回复我：Thank you for your submission of Manuscript ID Access-2020-32471 entitled "……” which you submitted to IEEE Access. IEEE Access requires that manuscripts are submitted as a PDF in a double column, single-spaced format using a mandatory IEEE Access template. Additionally, submissions should have the traditional IEEE format of an abstract, introduction, full explanation of background and history, a detailed explanation of recent findings, a conclusion, and a list of references. We have returned the article to your Author Center as a draft so that you can make the necessary corrections, and then resubmit. You may find additional information regarding submission and formatting requirements for IEEE Access at https://ieeeaccess.ieee.org/submitting-an-article/ Please also use this opportunity to carefully review and update the grammar before resubmitting since proper grammar is a requirement for publication in IEEE Access. When you are ready to resubmit, go to https://mc.manuscriptcentral.com/ieee-access and log in to your Author Center. Click on "Unsubmitted and Manuscripts in Draft," and then click on "Continue" located next to the manuscript number. Please be sure to make the necessary updates before following the steps to resubmit your manuscript. We look forward to your resubmission. Sincerely, IEEE Access Editorial Office.
求问怎么办