Voice Samples and Self-Incrimination

A Single Judge of the Gujarat High Court handed down the decision in Devani v. State of Gujarat [Spl. Criminal Application Direction 5226 of 2015, decided on 18.01.2017], which formed the basis of the previous Guest Post on the Blog. To recount, the decision agreed with the Supreme Court's conclusions in Ritesh Sinha [(2013) 2 SCC 357] that compelling a person accused of an offence to undergo 'Voice Spectrography' does not amount to making her a witness against herself. Having held so, Pardiwala, J. then considered whether there was any statutory basis to permit such testing by the police. This was the issue on which the two judges in Ritesh Sinha disagreed, and is now pending before a larger bench for resolution. Pardiwala, J. found no such basis, and thus held that the Petitioner could not be compelled to undergo Voice Spectrography. This brief post follows up on the analysis offered previously and develops those ideas further.

'Evidence' in Article 20(3) and the Physical-Mental Divide

To decide whether or not the right against self-incrimination, guaranteed under Article 20(3) of the Constitution, is violated, three questions need to be affirmatively answered. First, that the person concerned is accused of an offence. Second, this person is required to give evidence against herself. And, finally, that the person must be compelled to incriminate herself. In Devani, only the second of these questions arose: Pardiwala, J. had to decide whether the taking of voice samples/exemplars could amount to evidence? 

Section 3 of the Indian Evidence Act contains a very broad definition of the term 'evidence'. Does the same definition apply here? It doesn't. As Gautam pointed out, the interpretation of this part of Article 20(3) is governed by State of Bombay v Kathi Kalu Oghad [1962 (3) SCR 10] which was decided by eleven judges. Deciding whether or not handwriting samples and blood samples were within the pale of Article 20(3), the Supreme Court gave us the term 'physical evidence'. By prefacing 'evidence' in this manner, the Supreme Court excluded blood samples and handwriting samples from the mix. The logic essentially being, that these materials were inherently immutable and were only being used for drawing comparisons with material already collected during the investigation. Enter, Selvi v State of Karnataka [(2010) 7 SCC 263]. Decided by three judges, it considered whether an accused could be compelled to undergo either brain mapping (BEAP), polygraph tests, or narco-analysis. The three judges held that compelling a person accused to undergo any of these would violate the constitutional right under Article 20(3). Now, you may correctly ask, that was it proper to place all three tests in the same bracket? After all, the drugged state of narco-analysis involves persons making statements, as opposed to polygraph tests relying upon physical responses to stimuli (usually questioning). On the issue of polygraph tests, the Court held that such a physical response was as good as a statement. The response involved a unique communicative value, which ordinary 'physical evidence' would lack. 

I agree that Pardiwala, J. is not entirely correct in narrowly construing the contours of Selvi by suggesting that the decision was solely concerned with what he calls 'psychiatric evidence'. The problem really is how broad the conclusions in Selvi were, as Gautam pointed out. While it attempts to distinguish Kathi Kalu Oghad, the logic employed for bringing polygraph tests within the confines of Article 20(3) does not hold firm. This is because even in a polygraph test the physical responses are used to compare. There is a set of normal readings taken when the person responds to supposedly harmless questions. After this, the uncomfortable questions are asked, and the readings obtained by both the sets of responses are then compared. There is a way to keep both Kathi Kalu Oghad and Selvi, perhaps. This would look at how the preliminary readings in polygraph tests are also obtained under coercion, as against a case like blood samples where material is obtained wholly independently.

Self-Incrimination and Voice Exemplars

The previous post made important observations about the logic behind Article 20(3) which receives scant attention. Gautam argues that the employment of the 'physical evidence' concept was an attempt to balance the dual claims of the crime control vs. due process approach. Developing his thoughts further, one may suggest that Selvi shows a shift from focusing on the crime control aspects to the issues of due process. In Kathi Kalu Oghad, we get a clear idea that the Court saw Article 20(3) as a means to ensure the quality of evidence at trial remained of a sterling quality. While persons may give false testimony under pressure, blood never lies, and so it made no sense to exclude it from the possible evidence that a judge may consider to decide issues at trial. Fifty years later, the three judges in Selvi are at pains to link Article 21 to Article 20(3). There is as much attention placed upon the manner in which the evidence is obtained, rather than solely judging its quality. Thus, the Court appears more willing to look at issues such as particular techniques violating the mental privacy of the accused which it earlier would not rarely consider. 

The problem with making claims about 'shifts' is that there needs to be something more to suggest a trend. Unfortunately, I can offer nothing by way of evidence, and would rather classify Selvi as a one-off event in the largely depressing judicial history of Article 20(3). On each of the three questions that I identified, the Supreme Court has consistently narrowed the scope, only to render this constitutional protection non-existent for large swathes of the criminal law that are today administered by agencies other than the khaki-clad police.  

Moving on to considering the particular issue at hand, then. I agree with Gautam that the 'physical evidence' conception is not the answer for a conceptually rigorous model for understanding the right against self-incrimination. But I do not think that the issue of voice sample poses any great difficulty to the existing setup and thus calls for reform will not find many takers, yet. If we were to look at this issue of voice samples from the lens of Kathi Kalu Oghad, it would be argued that the voice sample is obtained to compare it with the tape recording. Voice, like handwriting, is quite immutable, making it a fit for the 'physical evidence' category. The approach offered by the Selvi would offer the same result. The voice samples here have no communicative quality of their own: they are purely for purposes of comparison with evidence that exists independent of the person. The big flaw in this entire claim is the idea that the human voice is immutable since there is material to suggest otherwise (something that Devani does not fully address). Assuming there is some certainty on that front, it is difficult to argue that the technique of Voice Spectrography would violate Article 20(3).

Creating a Statutory Basis

The second part of the decision is a familiar tale of woe. Pardiwala, J. traverses through the entire gamut of statutes in the field in his search for some basis to ground this technique, only to end his search in vain. He concludes that Voice Spectrography could therefore not be compelled upon an accused, and offers pointers to the State Government on how to remedy the situation before finishing his decision. The tale is familiar because it can be recounted for most investigative developments in India. This blog had earlier considered one example in Section 311-A of the Cr.P.C.; how it took several decades for such a common investigative tool (taking handwriting samples) to receive statutory support, only for it to be riddled with problems.

The Criminal Procedure Code, 1973 is a curious mixture of highly particular rules combined with vague standards. The part on police investigations are a good example. Police discretion on when to investigate a case is couched in vague terms, coexisting with provisions stipulating meticulous rules for things such as inquests. Do we, then, really need a clear statutory basis for what techniques can the police administer? Pardiwala, J. himself suggests that there may not be such a need when he invites the State Government to consider bringing in Rules to permit Voice Spectrography. Regardless of where one stands on the security-liberty paradigm, few would dispute the need for law to keep pace with technology and for the police to have up-to-date investigative tools at its disposal. This raises  questions on whether it makes sense to continue requiring concrete statutory backing for the particular techniques that the police can adopt, in an environment that is notorious for belaboured law reform. The spectre of abuse and misuse looms large though, leaving us with that familiar feeling: of being between a rock and a rather hard place. 

Guest Post: Voice Spectography and Self-Incrimination

(In this guest post, Mr. Gautam Bhatia offers his take on the recent decision by the Gujarat High Court in Devani v State, which considered the issues posed by using voice spectrography during investigations. This essay has been cross-posted with permission from the Indian Constitutional Law and Philosophy Blog) 

On the 18th of January, in Devani vs State of Gujarat, a single judge of the Gujarat High Court handed down an interesting ruling on the constitutionality and legality of voice spectrography as an investigative technique during criminal proceedings. The writ petitioner had been charged with offences under the Prevention of Corruption Act. The main piece of evidence was a telephone conversation between the petitioner and the person from whom he had allegedly demanded a bribe. The Investigating Agency wanted to subject the Petitioner to a voice spectrography test, so that it could compare the two voice samples. The Petitioner challenged this.

The Court’s decision was delivered in the context of a split opinion by a two-judge bench of the Supreme Court, delivered in 2012, in Ritesh Sinha vs State of U.P. Justice Aftab Alam had held that in the absence of express statutory authorisation, investigative agencies could not compel an accused to undergo a voice spectrography test. Justice Ranjana Desai had disagreed. While the two judges were in apparent agreement over the proposition that voice spectrography did not violate the right against self-incrimination under Article 20(3) of the Constitution, in view of the disagreement between them, they referred the case for resolution by a three-judge bench. This bench has not yet been constituted.

In Devani, the Gujarat High Court agreed with Justice Alam and disagreed with Justice Desai, holding that the investigating authorities could not legally compel an accused to undergo a voice spectrography test. The Court based its judgment on statutory interpretation, finding no warrant for the taking of voice samples under the existing criminal procedural framework. However, it also found that the taking of voice samples did not violate Article 20(3) of the Constitution. It is to this finding that we turn first.

Article 20(3)

The Court’s examination of the self-incrimination issue took place – as it had to – in the context of the Supreme Court’s 2010 decision in Selvi vs State (previously, on this blog, we have discussed Selvi and the right against self-incrimination under the Indian Constitution in some detail). Readers will recall that Selvi – a three-judge decision of the Supreme Court – had clarified some of the conceptual underpinnings of the right against self-incrimination that had been left open in the previous judgment of Kathi Kalu Oghad. The question in this case was how the Gujarat High Court would interpret Selvi, and how it would thread the needle between Kathi Kalu Oghad, which had held the taking of fingerprints and handwriting samples was consistent with Article 20(3), and Selvi, which had held that narco-analysis, brain mapping, and the polygraph test, were not. In other words, what was the deeper conceptual basis that underpinned the spectrum between fingerprints and narco-analysis, and where did voice spectrography fall on this spectrum?

The Gujarat High Court’s reading of Selvi was as follows: Selvi, it held, had prohibited the compulsory psychiatric examination of the accused. However, a voice spectrography test – which only involved the subject speaking something into a recorder, and subsequent analysis of his voice patterns – did not come within the ambit of “psychiatric examination“. Consequently, voice spectrography was not hit by the Selvi bar, and was therefore constitutional (paragraph 44).

It is submitted, with respect, that the High Court’s reading of Selvi is unnecessarily narrow, and is not borne out by the decision and reasoning in that case. Recall that, in the last analysis, in Selvi, the intellectual foundation of the Court’s judgment was the idea of mental privacy. In paragraph 160 of that judgment, the Court held:

“Even though the actual process of undergoing a polygraph examination or a BEAP test is not the same as that of making an oral or written statement, the consequences are similar. By making inferences from the results of these tests, the examiner is able to derive knowledge from the subject’s mind which otherwise would not have become available to the investigators. These two tests are different from medical examination and the analysis of bodily substances such as blood, semen and hair samples, since the test subject’s physiological responses are directly correlated to mental faculties. Through lie-detection or gauging a subject’s familiarity with the stimuli, personal knowledge is conveyed in respect of a relevant fact.” (Para 160)

On, in other words, the right against self-incrimination (among other things) was aimed at protecting the mental inviolability of an accused during the course of a criminal proceeding.

The Court in Selvi went on to note:

“The compulsory administration of the impugned tests impedes the subject’s right to choose between remaining silent and offering substantive information. The requirement of a `positive volitional act’ becomes irrelevant since the subject is compelled to convey personal knowledge irrespective of his/her own volition.”

And:

“While the ordinary exercise of police powers contemplates restraints of a physical nature such as the extraction of bodily substances and the use of reasonable force for subjecting a person to a medical examination, it is not viable to extend these police powers to the forcible extraction of testimonial responses. In conceptualising the `right to privacy’ we must highlight the distinction between privacy in a physical sense and the privacy of one’s mental processes… so far, the judicial understanding of privacy in our country has mostly stressed on the protection of the body and physical spaces from intrusive actions by the State. While the scheme of criminal procedure as well as evidence law mandates interference with physical privacy through statutory provisions that enable arrest, detention, search and seizure among others, the same cannot be the basis for compelling a person `to impart personal knowledge about a relevant fact’. The theory of interrelationship of rights mandates that the right against self-incrimination should also be read as a component of `personal liberty’ under Article 21. Hence, our understanding of the `right to privacy’ should account for its intersection with Article 20(3)… a conjunctive reading of Articles 20(3) and 21 of the Constitution along with the principles of evidence law leads us to a clear answer. We must recognise the importance of personal autonomy in aspects such as the choice between remaining silent and speaking.An individual’s decision to make a statement is the product of a private choice and there should be no scope for any other individual to interfere with such autonomy, especially in circumstances where the person faces exposure to criminal charges or penalties. Therefore, it is our considered opinion that subjecting a person to the impugned techniques in an involuntary manner violates the prescribed boundaries of privacy. Forcible interference with a person’s mental processes is not provided for under any statute and it most certainly comes into conflict with the right against self-incrimination.” (Paras 190 – 193)

Consequently, Selvi went much further than simply prohibiting compelled “psychiatric examinations”. It located its decision within the framework of personal autonomy, the right to remain silent, and mental privacy.

However, it is precisely within this context that the issue of voice spectrography attains an almost intractable difficulty. If you look at the issue from one perspective, it seems clear that voice spectrography falls squarely within the Selvi prohibition. The mind is centrally involved in the act of speaking, and if “mental privacy” is to be understood in this sense, compelling a person to speak, and then using his voice to determine other relevant facts in a criminal proceeding, is certainly an invasion of mental privacy. Compelling a person to speak certainly violates their choice between speaking and remaining silent. It is certainly an interference with her “mental processes“. And finally, it does seem that what happens in a voice spectrography test is very similar to polygraph tests and brain mapping – bodily stimuli are mapped and measured. The fact that in one case, it is stimuli from the brain, and in another, voice patterns, ought not to make a difference, because – as discussed above – speaking is a direct result of mental activity.

Examined another way, however, the issue is much more complex. While speaking is a result of brain activity, the crucial thing to note in a voice spectrography test is that it is not about what the speaker is saying – that is, not about the content of her voice – but about her voice as a bodily, physical phenomenon. In this sense, voice spectrography seems much closer to fingerprints, blood samples, and DNA, than it does to narco-analysis. In a similar sense, it seems a stretch to say that the examiner is deriving knowledge from the subject’s mind. Although speaking is a result of a mental process, given that the brain is, ultimately, responsible for keeping the human body alive, every other bodily function can, ultimately, be traced back to the brain. Here, again, given that the content of what the speaker says is not at issue, it does seem that voice samples are physical phenomena, in the same sense that blood or DNA is.

What this reveals, I suggest, is that ultimately, Selvi’s neat distinction between the physical and mental, between the taking of physical samples (blood, DNA, hair swabs, fingerprints) on the one hand, and interference with mental processes and invasion of mental privacy on the other, breaks down on closer examination. It breaks down in cases such as voice spectrography, where the physical/mental binary simply loses its valence as an explanatory device.

As I had argued in my essay on Selvi, the distinction between the physical and mental in that case was drawn in an attempt to strike a balance between two models of the criminal process: the crime-control model, which seeks to remove obstacles from investigating agencies in their attempt to discover the “truth”, and the due process model, which places great stress on the fundamental rights of the accused. By limiting the right against self-incrimination under Article 20(3) to invasions of mental privacy, the Court attempted to balance the two models. A case like Devani suggests that more work needs to be done to achieve a clear and coherent balance.

Statutory Analysis

Although not strictly within the scope of this blog, a word ought to be said about the second part of the judgment. The Court undertakes a thorough analysis of laws such as the Identification of Prisoners Act, the Code of Criminal Procedure, and the Evidence Act, all of which allow for various situations – and methods – by which investigating agencies and judges can act upon the body of the accused (compelling her to yield blood samples, DNA, sputum, hair swabs, fingerprints etc.) The Court finds that nowhere is there an express authorisation to take a voice sample, and nor can such an authorisation be read into the statutes. Consequently – and despite its misgivings on this issue as a matter of policy – the Court holds that, under the existing legal regime, compelled voice spectrography is not permissible. The Court’s analysis – from paragraphs 47 to 100 – repays close study, not simply because it is an excellent piece of statutory interpretation, but also because the Court subordinates its own sense of what might be desirable as a matter of legal policy to an autonomous reading of the existing law as it stands, and refuses to step an inch beyond the law. This is a phenomenon that has been falling distressingly out of fashion in recent years.