You are on page 1of 3

INTERNATIONAL CONFERENCE ON COMPUTATIONAL SCIENCES AND SUSTAINABLE

TECHNOLOGIES 2023
CHRIST (Deemed to be University), India and Modern College of Business and Science, Oman
ARTICLE REVIEW

Paper ID: 204

Paper Title: Forensic Speaker Verification Using GMM-UBM I Vector Method


in Intensive Background Noise

Review Questions:

1. Title and Abstract (Give detailed comment for suitability of Title and Quality of the
abstract)

The speaker verification experiment has been conducted by adding White


Gaussian Noise, Red Noise, and Pink noise with SNR ranging from -20 dB
to +20 dB for forensic applications. The performance of both methods got
affected drastically in call cases but autocorrelation based MFCC gave
better results than MFCC. Thus, autocorrelation based MFCC is a
valuable method for robust feature extraction when compared with
MFCC for speaker verification purposes in intense background noise.
The verification accuracy in this method is improved even in very high
noise level (-20dB) than the reported research work.

2. Literature Review / Background and Methods (Give detailed comments on the


understanding of the research work proposed / used and adequacy of literature
review)

The proposed work has been explained clearly with intention of detecting
the crime using speech recognition. The background of the study has been
well explained and it clearly states about the Dravidian language
Malayalam.

3. Result analysis and conclusion (Give detailed comments for quality of result analysis
and conclusion)

MFCC based autocorrelation based with noisy speech samples. It


produces excellent results with noiseless samples, but its performance
suffers when used with noisy data. The verification experiment was
carried out by incorporating White Gaussian Noise, Red Noise, and Pink
Noise with SNR ranging from -20 dB to +20 dB. Noise had a significant
impact on the performance of both methods in call cases, but
autocorrelation based MFCC significantly outperforms MFCC.

4. What are the strengths of this article? (Quality of research work like new methods
proposed, relevant to societal needs, review paper, case study etc.)

It is a good method to detect the crime using forensic speech recognition


in a noiseless environment.

5. What are the weaknesses of this article? (Basic study, poor organization of the
paper, other ineffective aspects)

It should also be compared with some other feature extraction methods


other than MFCC.

6. Overall Evaluation (Mention Points between 1 and 10 ) with Remarks

7. Confidential remarks for the program committee.

Paper is good with its quality

8. Decision
a. Accept
b. Accept with minor revision
c. Accept with major revision
d. Reject

If ‘Rejected’
Reason for rejection: ---

Reviewer’s Details:

Name: Dr.N.Revathy
(Enter as it should be printed in the Certificate)
Designation: Professor
Department: PG & Research Department of Computer Applications
Institution: Hindusthan College of Arts & Science (Autonomous)
Address: Hindusthan Gardens, Behind Nava India, Coimbatore – 641 028
Email: revathy.n@hicas.ac.in
Mobile: 98430 30579
Kindly affix Facsimile (if it is available) or digital signature

Note: Please send this form in pdf format

You might also like