---
title: "Why are some files marked as skipped?"
description: "How a batch run uses file size and modified time to decide a file is unchanged, why a skipped record produces no output, and how Skipped differs from Canceled"
canonical_url: "https://docs.ardge.com/en/knowledge/scribestt/skipped-files"
---
# Why are some files marked as skipped?

> How a batch run uses file size and modified time to decide a file is unchanged, why a skipped record produces no output, and how Skipped differs from Canceled

## Short answer

While scanning the source folder, if a batch run finds that the same task already transcribed a file successfully and the file hasn't changed since, it marks this record **Skipped** rather than running speech recognition again. That's deliberate: it avoids burning compute on re-transcribing unchanged files.

## Details

### How the decision is made

The system looks within one scope — same task, same source path — for an earlier transcription of that file with status **Completed**. If it finds one, it compares two pieces of metadata:

<table>
<thead>
  <tr>
    <th>
      Compared
    </th>
    
    <th>
      How
    </th>
  </tr>
</thead>

<tbody>
  <tr>
    <td>
      File size
    </td>
    
    <td>
      Exactly the same as the earlier record
    </td>
  </tr>
  
  <tr>
    <td>
      File modified time
    </td>
    
    <td>
      The same as the earlier record, to the second
    </td>
  </tr>
</tbody>
</table>

Both have to match for the file to count as unchanged and be marked **Skipped**. If either differs, it's treated as new content and transcribed again.

The scope is one task. When two different tasks scan the same file, each decides independently — neither skips because of the other.

### A skipped record produces no output

A **Skipped** record runs no speech recognition, so it produces no transcript and writes nothing to the target folder. The file that earlier successful transcription wrote is still in the target folder — there's no need to produce it again.

In **Run history**, these files count toward the **Skipped** column, not **Succeeded** or **Failed**.

### How Skipped differs from Canceled

Both are terminal states, but they arise differently and are handled differently:

<table>
<thead>
  <tr>
    <th>
      
    </th>
    
    <th>
      <strong>
        Skipped
      </strong>
    </th>
    
    <th>
      <strong>
        Canceled
      </strong>
    </th>
  </tr>
</thead>

<tbody>
  <tr>
    <td>
      When it happens
    </td>
    
    <td>
      At the scan stage, before processing starts
    </td>
    
    <td>
      Mid-processing, when it's aborted
    </td>
  </tr>
  
  <tr>
    <td>
      Did speech recognition run?
    </td>
    
    <td>
      No
    </td>
    
    <td>
      It started, then stopped
    </td>
  </tr>
  
  <tr>
    <td>
      Can it be retried?
    </td>
    
    <td>
      No. Retrying is pointless — the result already exists
    </td>
    
    <td>
      Yes
    </td>
  </tr>
</tbody>
</table>

## How to check

1. Go to the **Tasks** page
2. Click the run history icon on that task's row to open **Run history**
3. Look at that run's **Skipped** count
4. Click the row to open **Run details** and read each file's **Status** column

In **Run details**, a skipped file shows **Skipped** in the **Status** column and `-` in **Error message**.

To force a file to be transcribed again, change its content — which changes its size or modified time — or upload it on the **Transcriptions** page to create a separate transcription.

## Related articles

- [Connect a NAS data source and create a batch transcription task](https://docs.ardge.com/en/knowledge/scribestt/nas-batch-transcription)
- [What to do when a transcription fails](https://docs.ardge.com/en/knowledge/scribestt/transcription-failures)


## Sitemap

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