Skip to content
DraftTheRecordDraftTheRecord Docs

One-Time Offline Setup

Setup takes about 30 to 45 minutes and you do it once. Work through the steps in order, because each one depends on the one before it.

Watch The Walkthrough

Section titled “Watch The Walkthrough”

Every step below appears in the video in the same order.

What You Are Installing

Section titled “What You Are Installing”

DraftTheRecord App

The DraftTheRecord desktop app.

WSL

A Windows component Docker needs.

Docker Desktop

The app that runs the offline services.

Transcription Engine

The AI model that turns your audio into text, while running on your machine.

The Nine Steps

Section titled “The Nine Steps”

  1. Update DraftTheRecord

    Section titled “Update DraftTheRecord”

    Open DraftTheRecord and click Check for updates. Install anything it offers before going further.

    Clicking Check for updates in DraftTheRecord and letting the update install.
    Check for updates, then let it finish.

  2. Confirm Virtualization Is Enabled

    Section titled “Confirm Virtualization Is Enabled”

    Open Task Manager, go to the Performance tab, and click CPU. Look at the bottom right. It must say Virtualization: Enabled.

    Task Manager Performance tab showing CPU details, with Virtualization reading Enabled boxed in red in the lower right.
    Task Manager, Performance, CPU. Virtualization must read Enabled.

  3. Set Windows To Best Performance

    Section titled “Set Windows To Best Performance”

    Go to Settings → System → Power & battery → Power Mode. Set both Plugged in and On battery to Best Performance.

    Windows Settings Power and battery page with Power Mode being set to Best Performance.
    Both dropdowns set to Best Performance.

    Offline transcription is demanding. On a balanced or power saving profile, the transcript may lag behind the speaker.

  4. Update WSL

    Section titled “Update WSL”

    Open Windows PowerShell and run this:

    Windows PowerShell
    wsl --update
    Running wsl --update in Windows PowerShell and watching it complete.
    Let it finish before moving on.

  5. Install Docker Desktop

    Section titled “Install Docker Desktop”

    Go to docker.com/products/docker-desktop and choose Download for Windows, AMD64.

    Choose ARM64 only if you have a Surface Pro X. Almost all Windows laptops are AMD64.

    Downloading Docker Desktop for Windows AMD64 from docker.com.
    Download for Windows, AMD64.

    Then right-click the installer and choose Run as administrator.

    Right-clicking the Docker Desktop installer in the Downloads folder and choosing Run as administrator.
    Right-click, then Run as administrator.

    On the Configuration screen, select Per-user installation (Recommended) and tick Add shortcut to desktop.

    The Docker Desktop installer Configuration screen with Per-user installation selected and Add shortcut to desktop ticked.
    Per-user installation, and Add shortcut to desktop.

    When Docker opens, Accept the license, then skip the sign-in. You do not need a Docker account.

    Accepting the Docker Desktop license agreement and skipping the sign in prompt.
    Accept the license, then skip sign-in.

  6. Restart Your Computer

    Section titled “Restart Your Computer”

    This step is required. Docker will not work correctly until you restart.

    After restarting, open Docker Desktop and wait for Engine running in the bottom left corner.

    Docker Desktop with Engine running boxed in red in the bottom left corner.
    Wait for Engine running, bottom left, before continuing.

  7. Download The Transcription Engine

    Section titled “Download The Transcription Engine”

    Click the terminal icon (>_) in the bottom right of Docker Desktop. If Enable Docker terminal appears, click Enable.

    Clicking the terminal icon in the bottom right of Docker Desktop and enabling the terminal.
    Terminal icon, bottom right. Click Enable if prompted.

    Click Copy and paste into the terminal, then press Enter.

    Docker Desktop Terminal
    cmd /c "curl.exe -L --fail --ssl-no-revoke https://direct-uploads-sterling.s3.us-east-1.amazonaws.com/asr-usage-0.3.0.tar.gz | docker load && curl.exe -L --fail --ssl-no-revoke -C - -o sm-rt-en-15.0.0.tar.gz https://direct-uploads-sterling.s3.us-east-1.amazonaws.com/sm-rt-en-15.0.0.tar.gz && docker load -i sm-rt-en-15.0.0.tar.gz"
    The Docker Desktop Terminal downloading the transcription model with a time remaining counter.
    Progress with time remaining under Time Left.

    Even if you do not see anything loading on the screen, this step is still running in the background.

    You are done with this step when two “Loaded image:” lines print and the terminal returns to an empty prompt. Both lines are boxed in red below.

    The Docker Desktop Terminal at the end of the download, with both Loaded image lines boxed in red and the prompt back to empty.
    Both Loaded image lines, then an empty prompt.

  8. Start The Two Services

    Section titled “Start The Two Services”

    Part A - The Supporting Service

    Section titled “Part A - The Supporting Service”

    In the same terminal, paste this and press Enter.

    Docker Desktop Terminal
    docker run -v speechmatics-usage-data:/data -e EATS_PORT=9090 -p 19090:9090 -p 18000:8000 speechmaticspublic.azurecr.io/asr-usage:0.3.0

    You are looking for the line boxed in red below, reading Starting schedule: database purger. Leave the window open. This terminal now belongs to the supporting service, so it keeps printing and never returns to a prompt.

    The Docker Desktop Terminal running the supporting service, with the Starting schedule: database purger line boxed in red.
    Starting schedule: database purger. The service is up.

    Part B - The Transcription Service

    Section titled “Part B - The Transcription Service”

    This command is different for every machine, so you copy it directly out of the DraftTheRecord app.

    In DraftTheRecord, start a live offline session and open the Start Local Recording window. Under ALTERNATIVE COMMAND-LINE SETUP, copy the command shown by clicking the Copy button.

    Opening a live session in DraftTheRecord and reaching the Start Local Recording window.
    Live session, then Start Local Recording.
    Copying the machine specific command from under ALTERNATIVE COMMAND-LINE SETUP.
    Copy the command from ALTERNATIVE COMMAND-LINE SETUP.

    Open a new Docker Desktop Terminal window (+), paste your command and press Enter. Wait for it to report that it is listening.

    The Docker Desktop Terminal with the transport websocket listening line boxed in red.
    Transport websocket listening. The service is up.

  9. Test It

    Section titled “Test It”

    Back in the DraftTheRecord app, the Start Local Recording window should show the green message boxed in red below: Docker container is ready for offline transcription.

    The Start Local Recording window with the green message, Docker container is ready for offline transcription, boxed in red.
    Docker container is ready for offline transcription.

    Record 15 to 30 seconds of your own voice and confirm live text appears.

    Clicking Start Recording, confirming the green ready for offline transcription line in the dialog, and starting the recording.
    Start recording and speak normally.
    Live transcript text appearing on screen during the offline test recording.
    Text appears as you speak. Setup is complete.

From here on, the only page you need before a proceeding is What To Do Every Time.

Enhancing Or Assigning An Offline Job

Section titled “Enhancing Or Assigning An Offline Job”

An offline recording stays on your computer. Enhancing and assigning both happen in the cloud, so an offline job has to be uploaded first. It will not appear under Online until you do so.

  1. Leave The Job

    Section titled “Leave The Job”

    Come out of the job itself and go back to the sessions list.

  2. Press Upload

    Section titled “Press Upload”

    Find the session under Recent Live Sessions and press Upload on its row.

    Clicking Upload on the offline session in Recent Live Sessions, waiting for the upload to finish, then clicking Close.
    Press Upload on the session’s row.

  3. Wait A Few Minutes

    Section titled “Wait A Few Minutes”

    The upload takes a few minutes. When it finishes, the session appears in the Online tab.

  4. Enhance Or Assign

    Section titled “Enhance Or Assign”

    From the Online tab you can create an enhanced transcript or assign the recording to another organization, exactly as you would with a session recorded online.