CBI wants Jagdish Tytler’s voice samples in 1984 riots case: How voice testing works, its value as evidence

Senior forensic officials told The Indian Express that the frequency of a person’s voice remains the same for several years unless there is a medical condition in the voice chord or tract.

An official from Central Forensic Science Laboratory (CFSL), requesting anonymity, said, “A voice sample is generally taken in an echo-proof room for a controlled and noise-free environment and a voice recorder is used…the person is asked to speak a specific clue word from a statement already part of the evidence.

The official added that certain technical parameters are kept in mind while recording a person’s voice sample. “A spectral analysis of the audio speech is undertaken where the pitch, energy and frequency of the voice are kept as the base to study and match the original audio sample,” he said.

Forensic officials use international phonetic alphabets while recording a voice sample and ask the subject to pronounce only a small part of the original statement so that both vowels and consonants in the spoken bit can be alternatively analysed.

“There are two methods of comparison, you either get an anonymous voice sample and check it with a suspect list of five persons, or you know who the speaker is and corroborate both the voices,” said another official.

An official from Forensic Science Lab, Rohini, said the semi-automatic spectrographic method of voice sampling is used in Indian forensic labs while some countries use the automatic method where a likelihood ratio of the voice samples is developed, which increases accuracy.

The spectrographic method for speaker recognition makes use of an instrument that converts the speech signal into a visual display.

While the results of the voice sample either turn out to be positive or negative and the final report is submitted by the forensic lab to the investigating agency, officials said that inaccuracies mainly arise when the person’s voice is altered due to the effect of medicines, or if the person is suffering from a cold.

How does a matching voice sample help the police’s case in court?

Senior Delhi Police officers said that a matching voice helps confirm the evidence already collected by the investigating agency.

Retired Delhi Police officer Maxwell Pereira said: “Depending on the evidence, an investigating officer directs the forensic expert to extract a particular sample which will help their case in court. The credibility of the sample depends on the technique used by the expert and how the court analyses it for proving a person’s culpability.”

When was it first used?

The US Federal Bureau of Investigation (FBI) had first used the technique of voice identification analysis, also known as spectrographic at the time, as early as in the 1950s, but the procedure gained legitimacy in a 1962 study by Lawrence Kersta, a researcher working with a 1940s-model Bell Laboratory sound spectrograph.

Later, a research paper titled ‘Forensic Speech and Audio Analysis, Forensic Linguistics – A Review: 2001- 2004’, by APA Broeders, Chief Scientist, Netherlands Forensic Institute, said the parallel with the fingerprint that the voiceprint evoked, “was shown to be utterly misleading”.

The study, however, states that, “Unlike fingerprints, or friction ridge patterns, spectrographic representations of speech are not invariant over time but highly variable within speakers, reflecting the inherent within-speaker variability that is characteristic of speech”.

What were the past cases in India where voice samples were collected?

A special NDPS court had in February this year allowed a plea moved by the Narcotics Control Bureau (NCB), seeking collection of voice samples among 33 accused in a drugs case it was investigating after the death of actor Sushant Singh Rajput. NCB had claimed that it needed to verify certain voice calls that had emerged among the accused.

In December last year, the voice sample of Aftab Poonawala, accused of killing his live-in-partner and chopping her body into several pieces, was taken at CFSL after a local court allowed a police plea in light of an audio recording surfacing of the couple purportedly fighting.

In March this year, the Mumbai Police collected the voice samples of Aniksha Jaisinghani, daughter of alleged cricket bookie Anil Jaisinghani, accused of allegedly trying to blackmail and bribe Maharashtra Deputy Chief Minister Devendra Fadnavis’ wife, Amruta Fadnavis.

What is the legality behind collecting voice samples?

In a 2013 case, the Supreme Court considered whether compelling an accused to give his voice sample in the course of an investigation would be violative of the fundamental right against self-incrimination or the right to privacy.

Our criminal procedure laws do not contain a specific provision for testing voice samples because it is a relatively new technological tool. Collection of semen, hair samples for DNA analysis or taking general body measurements is routine and has specific provisions under law but for collection of voice samples, the police have to move court or seek consent of the accused.

Section 53 (1) of the Code of Criminal Procedure allows examination of accused by a medical practitioner at the request of a police officer. The provision reads: “When a person is arrested on a charge of committing an offence of such a nature and alleged to have been committed under such circumstances that there are reasonable grounds for believing that an examination of his person will afford evidence as to the commission of an offence, it shall be lawful for a registered medical practitioner, acting at the request of a police officer not below the rank of sub-inspector, and for any person acting in good faith in his aid and under his direction, to make such an examination of the person arrested as is reasonably necessary in order to ascertain the facts which may afford such evidence, and to use such force as is reasonably for that purpose.”

The word examination in this provision includes “the examination of blood, blood stains, semen, swabs in case of sexual offences, sputum and sweat, hair samples and fingernail clippings by the use of modern and scientific techniques including DNA profiling and such other tests which the registered medical practitioner thinks necessary in a particular case.”

Here, the phrase “such other tests” is read to include a collection of voice samples.

While delivering a split verdict in the 2013 case, the SC acknowledged that there is no specific law for collection of voice samples. The case was subsequently heard by a 3-judge bench in which the SC said that the fundamental rights of the accused will not be violated by collecting a voice sample for investigation.

The Court held that fundamental right to privacy cannot be construed as absolute and must bow down to compelling public interest.

In a ruling on March 30, 2022, the Punjab and Haryana High Court observed that “voice samples in a sense resemble fingerprints and handwriting, each person has a distinctive voice with characteristic features dictated by vocal cavities and articulates. The samples are collected after having permission in accordance with the law. The sample taken itself would not be an evidence, rather they are for comparing the evidence already collected.”