Training-Free Acoustic-Based Hand Gesture Tracking on Smart Speakers

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

Some notes re. review of Appisci-2678757 Training-free Acoustic-based Hand Gesture Tracking on Smart Speakers by  Xiao Xu  , Xuehan Zhang  , Zhongxu Bao  , Xiaojie Yu  , Yuqing Yin  , Xu Yang  , Qiang Niu

 

The 2 Preliminaries provides an informative introduction to the challenge, the technology and the proposed approach. But the adoption of a chirp signal could include some consideration of its spectral content and the influence of the choice of the ramp duration and frequency range . Also, as I read on through the paper, I encountered several points at which the performance of the system must be influenced by the repetition rate of the emission of the chirp pulses. But I cannot recall this being provided or discussed;; see for example line 205 where ∆t is the time increment..

In 3.2 Signal Enhancement  I understand that you create a wider chirp with the same frequency ramp and at a higher resolution. Are the conditions sufficient to ensure the pulse can be synchronized by shifting?

In 3.4.3 , how is the Kalman gain computed and did the multimodel based estimation system remain stable under all circumstances

The 4. Results section is very interesting and the system has been tested in a quite robust manner. The different environments would imply differences in reverberation but has the distance limitation to <2m negated such effects?

As a general comment – I note that many equation have a ‘,’ as part of the layout . In some cases as for example eqn(23) line217-218 the ‘,’ appears to change k-1 to k-1’

In the section  5.Discussion  the alternative method s for solving similar problems is a quite general review and since none of these have been used to compare the current work I think it is unnecessary at this point in the paper (whether an abbreviated version should appear in the Introduction could be considered)

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

The paper proposes a hand gesture recognition system based on signals of inaudible acoustic signals , designed to be device- and training free.

The acoustic signals are captured from the reflection of hand gestures from low cost speakers and microphones. Then, the hand position is obtained and bandpass filters are applied to cancel some interference and, lastly, the signals are enhanced purely on software.

The paper is well written and well structured, presents a quite interesting subject with an interesting idea. The main proposal seems more a technology innovation proposal than a research question. It is well applied to the topic of the journal, however, the presentation could be improved in scientific terms.

Some questions need to be answered to clarify and improve the results of the paper:

* How the bias of the volunteer knowing his hand gesture being recognized by the system could affect the results?

* The 5 gestures recognized by the system in the results are simple? How the system behave with complex gestures like Gestuno?

* Are some related work on literature? I suggest adding a related work section.

* What are the possible limitations of the recognition system? 

* What future directions the authors indicate to improve the recognition system?

Author Response

Please see the attachment.

Author Response File: Author Response.pdf