---
title: "ScribeSTT"
description: "Turn recordings and video into transcripts — set up the model, transcribe a file, improve terminology with a glossary, then batch-process a NAS folder on a schedule"
canonical_url: "https://docs.ardge.com/en/guide/ai-applications/scribestt"
---
# ScribeSTT

> Turn recordings and video into transcripts — set up the model, transcribe a file, improve terminology with a glossary, then batch-process a NAS folder on a schedule

## Overview

ScribeSTT turns meetings, interviews, recordings and video into text you can search, organise and hand to an AI.

An hour of meeting audio becomes a transcript with timestamps, which you can then use for meeting search, summaries, knowledge work, RAG or an AI agent. Once it's text, it's far easier to search, index, summarise and plug into the knowledge workflows you already have.

**Audio / video → transcript → search / summary / knowledge / AI**

Speech recognition runs in [AI Model Center](https://docs.ardge.com/en/guide/use-ai/ai-model-center) on the same ARX, so the recordings never leave the device.

![The ScribeSTT workspace overview](https://docs.ardge.com/screenshots/guide/en/scribestt.webp)

## Before you begin

Confirm the speech recognition model ScribeSTT needs is running in AI Model Center, and pick the right model instance in the app's settings — see [Set up the speech recognition model provider](https://docs.ardge.com/en/knowledge/scribestt/model-provider).

## Step 1: Upload one file, see a result fast

If you just want to try it, you don't need a NAS task first. Upload a piece of audio or video, choose the output formats, start the transcription, and watch the status.

When it finishes, read the transcript and its timing information — see [Upload a file and get a transcript](https://docs.ardge.com/en/knowledge/scribestt/upload-and-transcribe).

<note>

Before you upload anything, check [what formats and size limits ScribeSTT supports](https://docs.ardge.com/en/knowledge/scribestt/formats-and-limits) — 12 extensions, 1 GB per file, and 20 files per upload.

</note>

## Step 2: Improve proper nouns with a glossary

Company names, product names, technical terms and people's names are where a general speech model goes wrong most. Build a glossary that maps the misheard form to the correct one, and the whole transcript stays consistent.

See [Create a glossary](https://docs.ardge.com/en/knowledge/scribestt/glossary) for how.

## Step 3: Connect a NAS for batch processing

If your meeting recordings already sit on a NAS, register an SMB shared folder and let ScribeSTT batch-process it and write the results back.

**NAS folder → ScribeSTT task → transcript output folder**

See [Connect a NAS data source and create a batch task](https://docs.ardge.com/en/knowledge/scribestt/nas-batch-transcription).

## Step 4: Set a schedule

If new files arrive daily or weekly, have the task run at a fixed time so nobody has to trigger it — see [Schedule the batch task](https://docs.ardge.com/en/knowledge/scribestt/schedule-batch-runs).

## Step 5: Read success, failure and skipped

- **Success** — done
- **Failed** — find the reason before retrying, see [What to do when a transcription fails](https://docs.ardge.com/en/knowledge/scribestt/transcription-failures)
- **Skipped** — usually the file hasn't changed, so the system deliberately didn't process it again, see [Why are some files marked as skipped?](https://docs.ardge.com/en/knowledge/scribestt/skipped-files)

## Step 6: CLI, MCP and automation (advanced)

<note>

These are advanced integrations — most people can skip them at first.

</note>

- [Create a personal access token](https://docs.ardge.com/en/knowledge/scribestt/access-tokens) — the CLI and your own automation can't reuse your sign-in, so they authenticate with this instead
- [Wire transcription into your own tools](https://docs.ardge.com/en/knowledge/scribestt/mcp-cli-integration) — hand the tools to an AI agent over MCP, or call transcription from scripts and schedules with the CLI

## Which way should you use it?

<table>
<thead>
  <tr>
    <th>
      What you need
    </th>
    
    <th>
      Use
    </th>
  </tr>
</thead>

<tbody>
  <tr>
    <td>
      I have one meeting recording to transcribe
    </td>
    
    <td>
      Upload
    </td>
  </tr>
  
  <tr>
    <td>
      Product names keep coming out wrong
    </td>
    
    <td>
      Glossary
    </td>
  </tr>
  
  <tr>
    <td>
      I have a whole NAS folder
    </td>
    
    <td>
      Batch task
    </td>
  </tr>
  
  <tr>
    <td>
      I get new recordings every day
    </td>
    
    <td>
      Batch task + schedule
    </td>
  </tr>
  
  <tr>
    <td>
      I need to wire it into company automation or an agent
    </td>
    
    <td>
      CLI / MCP
    </td>
  </tr>
</tbody>
</table>


## Sitemap

See the full [sitemap](https://docs.ardge.com/sitemap.md) for all pages.