Evidence-grade transcripts and video analysis

Transcripts and footage review that hold up in court.

Affirma turns body-worn camera, interview and deposition recordings into speaker-labelled transcripts and event timelines. Every line links back to its source frame, and every export carries a verifiable record of how it was made.

See how attestation works

For law firms, courts, prosecutors, public defenders and law-enforcement agencies.

BWC-0417.mp4 · 1h 12m · Report AFR-2026-0193
Verified
REC 2026-10-14 22:47:13 · UNIT 4417
person 0.91
person 0.97
vehicle 0.94
00:0018:0036:0054:001:12:00
00:12:04 OFFICER 1

Step out of the vehicle for me, please.

00:12:09 SPEAKER 2

Why? What did I do?

00:13:51 OFFICER 1

You have the right to remain silent. Anything you say can be used against you…

Miranda advisement
sha256 9f3c2b1e … 04c8a71e · matches source Signed · Badge 4417

Who it’s for

Built for everyone who has to stand behind the record.

Whether you are building a case or defending one, Affirma gives every side the same verifiable account.

Law enforcement

For agencies, prosecutors and internal affairs

  • Draft incident narratives from body-worn and dash-camera footage, reviewed by the officer
  • Automatically detect use of force, Miranda advisements, arrests, searches and gunshots
  • Chain of custody from upload to a court-ready report
Affirma for law enforcement

How it works

From raw footage to a record you can attest to.

  1. 01

    Ingest

    Upload body-cam, dash-cam, interview or courtroom video. Each original is fingerprinted with SHA-256 the moment it arrives and never modified.

  2. 02

    Transcribe

    Speech is transcribed and separated by speaker, with a timestamp on every word, even across overlapping voices and street noise.

  3. 03

    Detect

    Models flag events in audio and video (use of force, threats, Miranda advisements, arrests, gunshots) with each pinned to its frame.

  4. 04

    Attest

    A reviewer corrects and signs off. The export bundles the transcript, event log and a manifest of every hash, model version and edit.

Video detection

Every event, pinned to the frame it came from.

Affirma watches and listens across hours of footage, then lays out what happened on a single timeline. Select any event to jump to the moment and read the transcript around it.

  • Initial contact
  • Threat
  • Use of force
  • Arrest
  • Miranda advisement
  • Search
  • Gunshot
BWC-0417.mp4 - detected events
11 events · 1h 12m
Video
Audio
Speech
00:0024:0048:001:12:00
Use of force 00:20:53 – 00:26:41 Jump to frame

Subject pulls away during handcuffing. Officer and subject go to the ground. Linked to 6 transcript lines and 2 camera angles.

signals v1.4 · confidence 0.93 Reviewed · Badge 4417
Certificate of record integrity Report AFR-2026-0193
Source
BWC-0417.mp4 · 1h 12m 06s
Source SHA-256
9f3c2b1e 77d0a5c4 … 6e1b04c8 a71e
Transcript SHA-256
41d09ab2 c37e0f18 … 92dd5e60 0b3f
Models
transcription v3.2 · signals v1.4
Edits
14 corrections by reviewer, each logged
Signed
2026-10-22 14:32 UTC · Badge 4417
Verified - every hash matches the original recording

Attestation

Show exactly how every transcript was made.

Machine output is only as credible as the record behind it. Affirma keeps that record for you, automatically.

  • Log every edit. See exactly what changed, when, and why.
  • Record the model and version behind every machine-generated line.
  • Export a signed manifest that anyone can check against the source.
Read about our security model

See Affirma on your own footage.

Bring a recording. We’ll return a transcript, an event timeline and an integrity certificate, so you can judge the results yourself.

Talk to our team