Skip to content
This is NOT an official site of the Government of Canada. Click here for the official AI registry.

AI-Assisted Audio Transcription and Translation for Investigators

Enforcement

What it collects that can identify you

Biometric
Identifiable data
  • Audio files containing voice recordings lawfully collected by the RCMP as evidence during investigations. These recordings may contain voices of identifiable individuals. The data is classified up to Protected B/Protected C.
Run by
Royal Canadian Mounted Police (RCMP)
Where
No fixed location
Kept
No data retained
Shared with
Accountable organization
Your copy
You cannot see the data it holds about you. What you can do

What it is for

This system uses AI speech recognition to automatically generate draft transcriptions and translations of audio files lawfully collected by the RCMP during investigations. Human transcribers and translators review and edit every output before it is used or disclosed. The system operates offline and on-premise; audio files and transcriptions are deleted from the system once the transcription has been downloaded.

What it collects and what happens to it

Data taken in

Biometric
Identifiable data
  • Audio files containing voice recordings lawfully collected by the RCMP as evidence during investigations. These recordings may contain voices of identifiable individuals. The data is classified up to Protected B/Protected C.

Processing

Speech & Audio
  • AI speech recognition converts audio files collected as investigative evidence into draft text transcriptions. The system also supports translation of audio content. The underlying AI engine was trained by a foreign government or non-government third party.

What it does

Sensing (Perceptive AI)
Human decides
  • The AI engine processes uploaded audio files and produces draft text transcriptions and/or translations. All outputs must be reviewed and edited by a human transcriber or translator before any use or disclosure; the raw AI output is never relied upon as-is.

Outputs

Generated content
Identifiable data
  • Draft text transcriptions and/or translations of the input audio files. These outputs may identify individuals whose voices appear in the recordings. They are draft documents requiring human review and editing before use or disclosure, and are deleted from the system after download.

Run by

Royal Canadian Mounted Police (RCMP)
  • The RCMP designed and operates the Voice2Text application within its Specialized Policing Services branch. The application is deployed offline and on-premise for use by RCMP transcribers and translators.

Voice2Text AIA

Built by

Not stated by the Helpful Places.

Kept for

No data retained
  • Once the transcription has been generated and downloaded by the user, the original audio file and the transcription are deleted from the system. No operational data is stored by the system long-term.
  • Duration: Deleted immediately after transcription is downloaded by the user

Shared with

Not available to me
  • Access to the system is restricted to authorized RCMP personnel via Active Directory login. Individuals whose voices appear in audio recordings do not have access to the transcriptions or to the system.
Available to the accountable organization
  • Transcriptions and associated audio are accessible to authorized RCMP transcribers, translators, and investigators during active processing. All data is deleted from the system after download.

Stored

Stored locally
  • The application is on-premise and offline; data is stored only on RCMP infrastructure within Canada for the duration of active processing. No third-party cloud storage is used.
  • Duration: Transiently, until transcription is downloaded and deleted
How to read the colours

Can it identify you?

Anonymized data
Data about people with the link to who is broken. Stripped of identifiers, blurred, aggregated, or noised so this system can’t reasonably tie a record back to an individual.
Pseudonymous data
Each person’s data is tied to a token (hash, ID, template) that lets this system recognise the same person across events, but the token itself doesn’t reveal a name. Reidentification is possible with extra information.
Identifiable data
The data either contains a direct identifier (name, address, account name, recognisable face or voice, plate number) or carries a token this system uses to look up legal identity during processing.

Who completes the loop?

Human decides
This mode suggests; a person decides what to do next. The AI is always advisory — a human is in the loop on every decision. Example: a triage tool ranks cases for a clinician who chooses which to see first.
Human executes
This mode decides; a person carries out the result. Example: an optimizer plans the day’s trash-collection routes, and drivers run them.
Autonomous
This mode decides and acts on its own. No person reviews each decision or carries out the resulting action.

Definitions from the DTPR standard. Amber is about your data, violet about who decides. The fuller the shape and the deeper the colour, the more identifying the data or the less a person is involved.

What you can do

Ask about this system

Questions go to the Helpful Places, not the vendor.

Your rights

  • Right to Be Informed of AI UseIndividuals whose audio is processed by this system are not directly notified of AI use at the point of processing. The AIA is published publicly on the Government of Canada Open Data portal, providing general notice that the RCMP uses AI-assisted transcription for lawfully collected investigative audio.
  • Right to Algorithmic TransparencyThe AIA for this system is publicly available on the Government of Canada Open Data portal. It describes the system's role as a transcription/translation assistant, the human-in-the-loop review requirement, input data types, and output format. The Directive on Automated Decision-Making requires a meaningful explanation of the system to be discoverable via a departmental website.

Risks and safeguards

  • Civil liberties harmAI speech recognition may produce inaccurate transcriptions that could affect the evidentiary record, potentially implicating individuals' due-process rights in investigations or court proceedings.Safeguard: All AI-generated transcriptions must be reviewed and edited by a human transcriber or translator before any use or disclosure. Raw AI output is never relied upon as-is. Audio classified up to Protected B/C is processed only on RCMP on-premise infrastructure, with access restricted to authorized personnel via Active Directory.