# Can I have some feedback on this research paper?

I have submitted this paper "Effect of Transducer Positioning in Active Noise Control" to arXiv.

I would like to get some feedback on the methodology of the same. Inputs of any sort are welcome. Can you please take a look at it and give me some feedback/suggestions/critics?

• While this would suggest an interesting use-case for SE, as a question, it is too broad. Do you think you could focus it a little bit? Are you looking for feedback on technical language? Topic? Methodology? – A_A Mar 1 '18 at 8:01
• So far, SE.DSP does not seem to be adapted to peer reviewing. But the answer-like principle could be a kind of open review. – Laurent Duval Mar 1 '18 at 8:16
• I would specially like to get comments on the methodology and results but inputs of all sorts are welcome. – SAJIL C K Mar 1 '18 at 8:19
• So, this might be the start of something different, or a complete disaster...so, depending how this goes, maybe...tell your friends (?) – A_A Mar 2 '18 at 15:07

Comments provided here are in two broad categories: Presentation and Subject matter. The "Presentation" section is the easiest to address. There are some things that could be rephrased in terms of language use but these might be just personal preferences. The "Subject matter" section includes comments in methodology which could take more time to address.

Overall, my perception was that the paper is an excellent start to more complex work but certain key points for this type of work were not clear enough. There is always "more work" that can be done but this particular paper has room for improvement before hitting those limits. I would give it another go before submitting to a journal.

1. Presentation

1. "Passive noise control uses bulky sound absorbent materials to mas unwanted sound, whose performance...", The control's or the material's?
2. "In the existing ANC literature, which has a history of 7 decades, the acoustic...", do we really need to load that phrase with yet another point on the way to saying something about the existing literature? It breaks the flow and I am not sure how much it adds to the point.
3. "This approximation, used for ease of computation, limits researchers....". Same comment applies here. Just say what you want to say in one phrase with two strong points rather than having one main point and an aside.
4. "...treatment of the research work is available at [some URL]...". Why not reference this properly and include one more citation? Or add a footnote if you think it would be better to have the URL immediately visible for some reason. As a reader, I am not even going to "parse" the URL unless I want to visit it and I have not yet read the message that this paper is trying to convey. Why send the reader away already?
5. "...using single or multiple microphones and passes [IT] on to the...","...THE TOOLBOX provided by Habets[20] HAVE been..." I am not going to flag each and every one of these tiny little things, please proof read the document. A typo can be seen a bit like "Oh come on man! I have worked my bottom off in this paper and you are picking on a typo!?!?!". But the thing is, the typo is brakng the flw. You are nt the redr of the papr, someone else is and they have no idea what you are trying to tell them. They "form" the idea by reading these lines, word-by-word. Typos get in the way. I stop, re-read the phrase, start again...It's more effort.
6. "ANC PRELIMINARIES" goes on for 1.5 paragraphs BEFORE revealing that what is described is actually depicted in Fig 2. For 1.5 paragraphs, I have been building this sketch up in my head reading the description. More effort on my part (the reader...not me personally).
7. "...Room Acoustics Models(RAM) are often used to recreate and reproduce room acoustics without any cross-model mismatch..." What is "cross-model mismatch"?
8. "...beamforming, acoustic auralization, speech.." Auralization is already acoustic.
9. "A RIR model channel [Fig 3]...". The figure's caption could have RIR captialised "Room Impulse Response...".
10. Are you sure you need II and III as separate sections? I don't know the audience of the paper, but maybe these two sections are at a kind of...very introductory level (?).
11. You start IV with "Figure 4 shows...", that's great. Why not do the same with the rest of the figure references?
12. Caption of Table 1 reads "Simulation settings used for an arbitrary case". We will come back to the "arbitrary case" later on in section 2 but as far as presentation of the work is concerned "ARBITRARY CASE" reads oddly. Perhaps this is the "TEST CASE" or "SIMULATION PARAMETERS"?
13. Add a grid to all of your graphs. It helps with reading the graph a lot.
14. "The source signal considered is sinusoidal with a wideband component", "We used the same method to generate the source signal as used here[22]...". Everything that is key to the results of this paper must be sufficiently mentioned in THIS paper. I (the reader...) am not going to go to your github repository to check your code. I may have a system of my own and I am just looking to improve or test it. I would really appreciate a simple formula there showing to me how was the "sinusoidal with a wideband component" created.
2. Subject Matter

1. Section IV:

1. "The room size was chosen arbitrarily to a size of...". Why arbitrarily? What is wrong with "The average auditorium size is...therefore we focused on rooms of that size", or, "In our research we chose to focus on rooms of size BLAH because that's the standard listening room size". Maybe you can refer to some sort of authority that is sending standards for room sizes. Maybe the equipment you develop is ideal for rooms of BLAH size? In any case, for me "...chosen arbitrarily..." is not perceived positively. The point is why did you choose that room size?
2. "The z level of [THE] microphone, noise source and microphone were fixed to a height of 1.53..." WHY?. First of all, 1.53 (what? Probably meters) and then WHY?. Maybe that's the level an average person's ear ends up when they are sitting? Otherwise...WHY 1.53 meters? Is that the average size person? Is there any other reason?
3. "...for ease of computation....". No.
4. "The x coordinate[S] [...] were chosen randomly...". Why randomly? What is the reason? I don't know if it is clear what I am saying here. I am not trying to put you on the spot. What I am reading is "We took some tools, we threw some dice, we got these numbers"...Which, I don't think is what you are trying to say here. Pin these parameters down because of a reason. Then the paper gets some perspective with reality.
5. "The antinoise source positions were evenly spaced at a distance of 11cm". WHY?.
6. "Table 1 shows an arbitrary case..."... Shows one of the test cases considered, see previous comment... There is too much "arbitrary", "random".
7. *"The sampling frequency (Fs) is fixed to 2kHz"....WHY?. For example, you could say something like, "Given the current performance of DSP / speakers, realistic ANC is only possible for frequencies between BLAH and BLAH Hz. We targeted this band and therefore worked with Fs=2kHz.".
8. "...according to Nyquist theorem...". No.
9. *"The adaptive filter [ORDER] was fixed [AT] 350 [COEFFICIENTS] at Fs=2kHz which is an important parameter determining ANC performance."... WHY?. Why is 350 at 2kHz important for performance? Are you talking about computational performance or whether the ANC does its job?
10. "A step size of .." It is not entirely clear that you "sweeped" the locations of the mics and the speakers.
2. THE POINT ABOUT FIR FILTERS: In the introduction it is stated "In the existing ANC literature, [...], the acoustic channels are mostly modelled using [FIR] filters. This approximation, [...], LIMITS researchers in studying the effects of physical aspects like transducer positioning, orientation, room reverberation, wall reflections, room structure etc. [...] RIR is analogous to AN IMPULSE RESPONSE in LTI systems. The IR is considered to characterise the behaviour of an LTI system completely.....". Given that you are using FIR filters everywhere in the paper...Shouldn't this part be rephrased / removed completely?

1. The 350 tap LMS is FIR
2. RIRs are FIR.
3. Equations (1) and (2) are FIR
3. THE NATURAL ROOM FILTER. A room is already a filter. When you pump sound in the room, as soon as you have reflections, you get interference. Interference puts a grid of "nodal points" in the room. There are certain locations in the room that if you put a microphone there, naturally, you will be recording a very low signal. This is the reason behind positioning speakers at certain distances from the walls so that you have the waves projected at the back of the speaker work to your advantage. I am looking at your Fig 9 and I am thinking that I can almost see the grid in the Y direction. The point is: Given a room, a mic and a speaker, the mic might be recording reduced sound levels because of its natural positioning. WITHOUT ANC. The room itself is filtering. Fig 9's caption says "...It can [BE] seen that the good anti-noise source positions..."...perhaps these are not "good", but "natural" (?). Because of the physics...That's where you expect them to be. Not because of the ANC. So, what you are saying is, I am taking into account the natural acoustics of the room to correct for what the signal sounds like at the point of the mic knowing the room it propagates and this helps in ANC (?)...

4. Fig 11 shows that the ANC is working. This is an ANC, clearly targeting the band $(0 \ldots \approx 1000Hz]$. What is its effect on useful signal that occupies the same region?

5. SIMULATION METHOD / NOVELTY:. Figure 3 implies that you are using the ray tracing model of sound propagation, which, for a rectangular empty enclosed space is fine. But we are not in the 70s. So, in the presence of this fantastic (and FOSS) piece of software, maybe you could use impulse responses from realistic (and large) auditoriums with specific constraints (?). In i-Simpa you can control everything...And it's free...Which means that you can download it now and start playing.

Hope this helps.

• wow! i've reviewed a few papers for the Journal of the AES, but doing so for an arXiv paper is a lot of work. i hope the OP appreciate this, A. – robert bristow-johnson Mar 3 '18 at 0:13
• Thank you so much @A_A for spending time for this and giving such detailed feedback. This is exactly what I wanted, and I really appreciate your work. – SAJIL C K Mar 4 '18 at 2:47
• @robertbristow-johnson Thanks, I think that arxiv would be even more lenient than this because it is a pre-print server. I reviewed it with the criterion of "how close is it to publication / have we done the best we could before sending it off?" and because I wanted to see the experiment advancing. – A_A Mar 5 '18 at 9:10
• @SAJILCK Thank you for letting me know. I am glad to hear it was useful. With the opportunity of these comments, I scanned the post again and found that for a post that makes a fuss about typos, it contained quite a few of them but in any case, I hope the point came across clearly. Please see previous comment to RBW about how this post triggered an "experiment" which I thought was interesting and wanted to participate. All the best with your work. – A_A Mar 5 '18 at 9:15
• A_A: Thanks for your time. I'll leave the peer-review tag in and let's see if others come up. As you say on Meta, it might be better to do this via Arxiv, but this seemed to work OK. – Peter K. Mar 5 '18 at 13:06