Multimedia Forensics: Speaker Identification and Selvi 2010 Admissibility
Published:
Reviewed by Bismith B · 07 Jun 2026
Questions
30
Duration
30 min
Faculty-reviewed
0
Updated
26 May 2026
About this mock
This mock test covers two interlocking pillars of Unit VIII of the UGC-NET Forensic Science Paper II syllabus: the science of speaker comparison and the Indian legal framework for admitting voice recordings in court. Questions probe the aural-spectrographic method (Tosi 1971 MSU study, narrow-band vs wide-band spectrograms, NAS 1979 critique), the evolution of automatic speaker recognition from GMM-UBM (Reynolds 2000) through i-vector (Dehak 2011) and x-vector (Snyder 2018) to deep speaker embeddings (d-vector), closed-set vs open-set identification tasks, the 1:N vs 1:1 distinction, the Bayesian likelihood ratio (LR) framework per ENFSI and IAFPA guidelines, and forensic casework involving tapped calls, ransom calls, and threat calls intercepted under Section 5(2) of the Indian Telegraph Act, 1885.
The Indian law strand runs through the real judgments that govern this area: Selvi v State of Karnataka (2010) 7 SCC 263 on Article 20(3) and Article 21 bars to involuntary testimonial compulsion; Ritesh Sinha v State of UP (2019) 8 SCC 1 on the Constitution Bench clarification of magistrate power to compel voice samples; State of Bombay v Kathi Kalu Oghad (1961) 3 SCR 10 on the testimonial vs non-testimonial distinction; Anvar P.V. v P.K. Basheer (2014) 10 SCC 473 on the mandatory Section 65B certificate (now Section 63 BSA 2023); and Arjun Panditrao Khotkar (2020) 7 SCC 1 overruling the Shafhi Mohammad relaxation. Expert admissibility under Section 45 IEA 1872 (now Section 39 BSA 2023), CFSL audio-forensics practice at Hyderabad, Chandigarh, and Kolkata, and the Daubert vs Frye contrast for comparative context round out the coverage.
Topics covered:
- Tosi 1971 aural-spectrographic method, NAS 1979 critique, narrow-band vs wide-band spectrograms
- GMM-UBM (Reynolds 2000), i-vector (Dehak 2011), x-vector (Snyder 2018), d-vector deep embeddings
- MFCCs and vocal tract encoding; closed-set vs open-set ID; speaker ID (1:N) vs verification (1:1)
- Bayesian LR framework: P(E|Hp)/P(E|Hd); ENFSI verbal equivalence scale; IAFPA validation gate
- Selvi v Karnataka (2010): Article 20(3)+21, testimonial vs non-testimonial compulsion
- Ritesh Sinha v UP (2019): magistrate implied power; voice sample as non-testimonial
- Kathi Kalu Oghad (1961), Anvar P.V. (2014), Arjun Panditrao (2020): Section 63 BSA certificate
- Section 63 BSA 2023 / IEA 65B certificate; Telegraph Act Section 5(2); Daubert vs Frye
Allow 30 minutes.
Sources & references
Questions in this mock are written and verified against the following sources. Citations are recorded per question and shown in the explanation after submission.
- cited in 6 questions
Maher, Robert C. -- Principles of Forensic Audio Analysis, Springer, 2018
Chapter 6: Automatic Speaker Recognition, TDNN-based x-vector systems
- cited in 4 questions
IAFPA -- Code of Practice and Code of Ethics, International Association for Forensic Phonetics and Acoustics
Section 5: Acceptable Methodologies and Reporting Standards, Aural-Spectrographic and LR Approaches
- cited in 4 questions
Rose, Philip -- Forensic Speaker Identification, CRC Press, 2002
Chapter 2: History and Validity of Spectrographic Speaker Identification, Tosi 1972 JASA study
- cited in 2 questions
Bharatiya Sakshya Adhiniyam, 2023
Section 39: Opinion of Experts (replacing IEA Section 45); qualification as expert by special skill
Open source - cited in 2 questions
Selvi v State of Karnataka, (2010) 7 SCC 263
Supreme Court of India, distinction between testimonial and non-testimonial compulsion as applied to voice samples
- cited in 2 questions
Hollien, Harry -- The Acoustics of Crime: The New Science of Forensic Phonetics, Plenum Press, 1990
Chapter 5: Spectrographic Analysis, Narrow-Band vs Wide-Band Filter Roles
- cited in 2 questions
ENFSI -- Best Practice Manual for Forensic Speech and Audio Analysis, European Network of Forensic Science Institutes
Section 5: Reporting, Verbal Equivalence Scale for Likelihood Ratio Conclusions
- cited in 1 question
Bharatiya Sakshya Adhiniyam, 2023 and Arjun Panditrao Khotkar v Kailash Kushanrao Gorantyal, (2020) 7 SCC 1
Section 63 BSA 2023 certificate requirements for electronic records; application to mobile-retrieved recordings
Open source - cited in 1 question
Ritesh Sinha v State of Uttar Pradesh, (2019) 8 SCC 1
Supreme Court Constitution Bench on magistrate power to compel voice sample and Article 20(3) analysis
- cited in 1 question
Arjun Panditrao Khotkar v Kailash Kushanrao Gorantyal, (2020) 7 SCC 1
Supreme Court on Section 65B certificate mandatory requirement, overruling Shafhi Mohammad (2018)
- cited in 1 question
State of Bombay v Kathi Kalu Oghad, (1961) 3 SCR 10
Supreme Court eleven-judge bench on Article 20(3) scope: testimonial vs non-testimonial compulsion
- cited in 1 question
Reynolds, D.A., Quatieri, T.F., Dunn, R.B. -- Speaker Verification Using Adapted Gaussian Mixture Models, Digital Signal Processing, 2000
Section 3: UBM Architecture and MAP Adaptation for Speaker Enrollment
- cited in 1 question
Anvar P.V. v P.K. Basheer, (2014) 10 SCC 473
Supreme Court on mandatory Section 65B(4) IEA certificate for electronic records; applicable to call recordings
- cited in 1 question
Dehak, N. et al. -- Front-End Factor Analysis for Speaker Verification, IEEE Trans. Audio Speech Lang. Process., 2011
Section 2: Total Variability Space and i-Vector Extraction
- cited in 1 question
Indian Telegraph Act, 1885 and Bharatiya Sakshya Adhiniyam, 2023
Section 5(2) Telegraph Act interception; Section 63 BSA 2023 certificate requirement per Anvar P.V. (2014) 10 SCC 473
How our mocks are built
Questions are written and edited by the ForensicSpot team and cited from peer-reviewed forensic textbooks, official syllabi and primary case law. Each one is verified before publishing. Detailed explanations show after you submit, so the test stays a real test. See a mistake? Tell us.
Common questions
What does the Multimedia Forensics: Speaker Identification and Selvi 2010 Admissibility mock cover?+
This mock test covers two interlocking pillars of Unit VIII of the UGC-NET Forensic Science Paper II syllabus: the science of speaker comparison and the Indian legal framework for admitting voice recordings in court. Questions probe the aural-spectrographic method (Tosi 1971 MSU study, narrow-band vs wide-band spectrograms, NAS 1979 critique), the evolution of automatic speaker recognition from GMM-UBM (Reynolds 2000) through i-vector (Dehak 2011) and x-vector (Snyder 2018) to deep speaker embed
How many questions and how long is the test?+
30 multiple-choice questions, 30 minutes total. Difficulty: hard. Tier: Premium.
Who is this mock for?+
Forensic science students and aspirants who want timed, exam-style practice with explanations and verified source citations on Multimedia Authentication and Deepfake Forensics, NET. Useful for postgraduate entrance preparation and for BSc / MSc forensic students testing their recall under time.
Are the questions reviewed?+
Each question carries a verified source citation. Faculty review for individual questions is in progress.
Do I need an account to take this mock?+
Yes, a free ForensicSpot account is required to start a timed attempt — this lets you save progress, see per-question explanations after submission, and track your topic-level performance over time.