Audio and Video Recordings as Evidence under the Bharatiya Sakshya Adhiniyam, 2023: The Conditions in Ram Singh, Voice Identification and Proof

Recordings of conversations have been received in Indian courts since the 1960s, and the conditions on which they are received were settled long before the law of electronic evidence existed. Those conditions have not been displaced. A recording tendered today must satisfy both the old judicial requirements — voice identification, accuracy, absence of tampering, safe custody, audibility — and the statutory requirements of Sections 57, 61 and 63. Satisfying one and not the other is of no use at all.

1. A Recording Is a Document

A recording is matter recorded on a substance by means capable of recording it, and it is a document within Section 2(1)(d). The illustrations to that definition name voice mail messages stored on digital devices expressly, and an audio or video file falls within 'electronic and digital records' in any event.

Two consequences follow that are easily overlooked. A recording is proved as a document, not by calling the person who made it and asking what was said — his testimony is evidence about the recording, and the recording is the evidence of the conversation. And the whole of Part III applies, including the rules about primary and secondary evidence and about how much of a record must be produced.

2. The Foundational Decisions

Three decisions established the position and remain the starting point.

3. The Six Conditions Examined

3.1 Identification of the voice

The recording establishes that words were spoken; it does not establish by whom. Identification proceeds by three routes and they are not equally strong.

Recognition by a person familiar with the voice is the ordinary route. A witness who knows the speaker deposes that the voice is his, and is cross-examined on the basis and extent of his familiarity and on the quality of the recording.

Admission by the speaker is the strongest, and is often obtained in cross-examination where the content is not damaging in itself.

Expert comparison with a voice sample is available under Section 39(1), whose residuary words now accommodate voice analysis without any argument about whether it is a 'science'. The court's power to compel a sample is settled.

3.2 Accuracy

The maker of the recording must prove that it accurately reproduces the conversation. What is required is evidence of who made the recording, with what device, when, where, and in what circumstances, and that the recording covers the whole of the relevant conversation.

A recording produced by somebody who cannot say how it was made, or who was not present when it was made, has not been proved accurate, and the condition is not satisfied by the recording sounding plausible.

3.3 Absence of tampering

Ram Singh places the burden on the party tendering the recording to rule out the possibility of tampering or erasure. This is a demanding requirement and it is the condition that most often fails.

Audio and video are edited without difficulty and without visible trace, and a party cannot discharge the burden simply by asserting that no editing occurred. What discharges it is preservation of the original recording in its original device or format, a hash computed at the earliest moment, sealing and custody, and where the point is contested, expert examination of the file for discontinuities, splices, encoding inconsistencies and metadata anomalies.

3.4 Relevance

The recording must be relevant under the ordinary heads. Most commonly it is relevant as part of the transaction under Section 4, on the reasoning in R.M. Malkani; as an admission under Section 15 where the speaker is a party; as conduct under Section 6(2); or as showing a state of mind under Section 12.

Where the speaker is not a party and the recording is offered to prove the truth of what he said, it is hearsay, and proving the recording under Section 63 does not make the assertion true.

3.5 Sealing and custody

The recording must be sealed and kept in safe or official custody. This requirement predates the modern chain-of-custody discipline and expresses the same idea: the court must be able to see that what it is hearing is what was recorded.

3.6 Audibility

The voice must be clearly audible and not lost or distorted. A recording in which the material words cannot be heard proves nothing, and a transcript prepared by an interested party does not cure it — the transcript is an aid to listening, not a substitute for it, and where the parties differ about what is audible the court must listen for itself.

4. The Standard Applied

This decision is the modern statement of why the Ram Singh conditions are applied strictly. The reasoning has, if anything, strengthened: editing tools that once required equipment and skill are now available to anybody, and the absence of visible signs of editing establishes nothing.

5. The Statutory Layer

The judicial conditions govern weight and authenticity. The statutory provisions govern admissibility, and both must be satisfied.

Primary or secondary? The Explanations to Section 57 apply. Explanation 3 makes each of several files storing a record primary evidence. Explanation 4 makes a record from proper custody primary evidence unless disputed — which, in a case about a recording, it invariably will be. Explanation 5 is directly in point where a video recording was simultaneously stored and transmitted or broadcast, and it makes each stored recording primary evidence without the qualification that weakens Explanation 4.

Where the certificate is required, the four conditions in Section 63(2) must be satisfied and the certificate under Section 63(4) furnished in the form of the Schedule, with the hash value in Part A and an expert declaration in Part B.

6. Transcripts

A transcript is almost always produced with a recording, and its status should be clear.

A transcript is not the evidence. The recording is the evidence, and the transcript is an aid to following it. It is prepared by somebody, and that somebody's accuracy, competence and impartiality are all open to question.

Three points follow. The person who prepared the transcript should be identified and, where it is contested, examined. Where the parties differ about a passage, the court must listen to the recording and form its own view. And where the conversation is in a language the court does not follow, a translation is a further layer requiring its own proof — the translator being, in effect, an expert whose competence must be established.

7. Completeness

A recording of a conversation engages the section on two footings, being both a conversation and an electronic record. The requirement is the same as with any exchange: enough must be produced for the court to understand what was said and the circumstances in which it was said.

A party producing an extract from a longer recording, or a recording that begins part-way through a conversation, should expect to be asked for the rest. Where it is withheld, Illustration (g) to Section 119 permits the presumption that what is withheld would have been unfavourable — and in the case of a recording the inference is a natural one, since a party who recorded a conversation ordinarily has the whole of it.

8. Recordings Made Without Consent

A recording made without the knowledge of one participant raises three questions, and the answers are settled.

Is it admissible? Yes. Yusufalli Nagree and R.M. Malkani both concerned surreptitious recordings and both held them admissible. The general Indian position, established in Pooran Mal v. Director of Inspection (Investigation), (1974) 1 SCC 345, is that the test of admissibility is relevancy, and evidence is not excluded because of the manner in which it was obtained unless a statute so provides.

Does it offend Article 20(3)? No. The protection is against being compelled to be a witness against oneself. A person who speaks of his own volition, unaware that he is being recorded, is not compelled, and both decisions so held.

Is it lawful? That is a different question from admissibility. Interception of communications is regulated by the Information Technology Act, 2000 and by the telegraph legislation, and unauthorised interception may be an offence. A recording obtained unlawfully may nevertheless be admissible on the Pooran Mal principle, while exposing the person who made it to consequences — and the circumstances in which it was obtained remain a legitimate subject of comment on its weight.

9. A Practical Checklist

1. Preserve the original device or file, not a copy shared through a messaging application, which re-encodes and strips metadata.
2. Compute a hash at the moment of preservation and record it contemporaneously.
3. Seal and keep the record in safe custody, as Ram Singh requires.
4. Identify the maker and be ready to prove when, where, with what device and in what circumstances the recording was made.
5. Plan the voice identification — a witness familiar with the voice, an admission, or an expert comparison with a sample.
6. Produce the whole recording, not an extract, because Section 33 requires it and withholding invites an adverse inference.
7. Prepare the transcript carefully, identify who made it, and treat it as an aid rather than as the evidence.
8. Obtain the certificate under Section 63(4), and consider whether Explanation 5 to Section 57 applies where the recording was simultaneously stored and transmitted.

10. The Position Stated Shortly

1. A recording is a document and is proved as one, not by the testimony of the person who made it.
2. Recordings have been admissible since Yusufalli Nagree, and a contemporaneous recording is relevant as part of the transaction — R.M. Malkani.
3. The six conditions govern — laid down in Ram Singh — voice identification, accuracy, absence of tampering, relevance, sealing and custody, and audibility.
4. The burden of ruling out tampering lies on the party tendering, and the standard of proof as to authenticity is more stringent than for ordinary documents — Tukaram Dighole.
5. The statutory layer applies as well. Explanations 4 and 5 to Section 57, and where they do not apply, Section 63 and the certificate.
6. The hash value is the modern answer to the anti-tampering condition, and it is far stronger than oral assurance.
7. A transcript is an aid, not the evidence, and where the parties differ the court must listen for itself.
8. A surreptitious recording is admissible and does not offend Article 20(3), though the circumstances of making bear on its weight.

11. Related Topics and Provisions