---
title: "What to do when a transcription fails"
description: "Where to find the failure reason, the common failure messages and how to handle each, and which statuses can be retried or canceled"
canonical_url: "https://docs.ardge.com/en/knowledge/scribestt/transcription-failures"
---
# What to do when a transcription fails

> Where to find the failure reason, the common failure messages and how to handle each, and which statuses can be retried or canceled

## Short answer

When a transcription's status reads **Failed**, open that record's **Transcription details** for the failure reason, clear the cause, then retry the record. Retrying without clearing the cause gets you the same result.

## Details

### Where to find the failure reason

- **Transcriptions created by upload** — on the **Transcriptions** page, click the file name to open **Transcription details**. It shows "Transcription failed" with the reason below
- **Transcriptions from a batch task** — on the **Tasks** page, open that run's **Run details** and read the **Error message** column on the file's row

### Common failure reasons and what to do

<table>
<thead>
  <tr>
    <th>
      Message
    </th>
    
    <th>
      Likely cause
    </th>
    
    <th>
      What to do
    </th>
  </tr>
</thead>

<tbody>
  <tr>
    <td>
      <code>
        The speech recognition model isn't ready or isn't deployed. Confirm the model is running and try again.
      </code>
    </td>
    
    <td>
      The model isn't running right now, or the model named on <strong>
        Model connection
      </strong>
      
       no longer exists
    </td>
    
    <td>
      Ask an administrator to check the model status — see <a href="https://docs.ardge.com/en/knowledge/scribestt/model-provider">
        Set up the speech recognition model provider in ScribeSTT
      </a>
    </td>
  </tr>
  
  <tr>
    <td>
      <code>
        Can't reach the ASR model provider — check the base URL and the network.
      </code>
    </td>
    
    <td>
      The <strong>
        Base URL
      </strong>
      
       on <strong>
        Model connection
      </strong>
      
       is wrong, or the network to the model service is down
    </td>
    
    <td>
      Ask an administrator to verify the <strong>
        Base URL
      </strong>
      
       and the network
    </td>
  </tr>
  
  <tr>
    <td>
      <code>
        Unexpected response format from the ASR model provider.
      </code>
    </td>
    
    <td>
      The bound endpoint isn't OpenAI-compatible, or the <code>
        /v1
      </code>
      
       version prefix is missing
    </td>
    
    <td>
      Ask an administrator to fix the <strong>
        Base URL
      </strong>
    </td>
  </tr>
  
  <tr>
    <td>
      <code>
        This file has no transcribable audio
      </code>
    </td>
    
    <td>
      The video file has no audio track
    </td>
    
    <td>
      Use a file that has one
    </td>
  </tr>
  
  <tr>
    <td>
      <code>
        The audio file couldn't be decoded — it may be corrupt
      </code>
    </td>
    
    <td>
      The source file is damaged
    </td>
    
    <td>
      Regenerate or re-obtain the file
    </td>
  </tr>
  
  <tr>
    <td>
      <code>
        Unsupported file format
      </code>
    </td>
    
    <td>
      The extension isn't on the supported list
    </td>
    
    <td>
      See <a href="https://docs.ardge.com/en/knowledge/scribestt/formats-and-limits">
        What file formats and size limits does ScribeSTT support?
      </a>
    </td>
  </tr>
  
  <tr>
    <td>
      <code>
        File too large
      </code>
    </td>
    
    <td>
      The file exceeds 1 GB
    </td>
    
    <td>
      Split the file and upload again
    </td>
  </tr>
  
  <tr>
    <td>
      <code>
        SMB connection failed
      </code>
    </td>
    
    <td>
      The data source's connection settings are wrong, or the NAS is unreachable
    </td>
    
    <td>
      Run <strong>
        Test connection
      </strong>
      
       again on the <strong>
        Remote data sources
      </strong>
      
       page
    </td>
  </tr>
  
  <tr>
    <td>
      <code>
        Insufficient NAS permissions
      </code>
    </td>
    
    <td>
      The data source's account lacks permissions on that folder
    </td>
    
    <td>
      Adjust that account's access on the NAS
    </td>
  </tr>
  
  <tr>
    <td>
      <code>
        The transcription result is missing timeline information. Retry.
      </code>
    </td>
    
    <td>
      The speech recognition response had no timeline, so SRT couldn't be produced
    </td>
    
    <td>
      Retry directly
    </td>
  </tr>
  
  <tr>
    <td>
      <code>
        The audio probe couldn't run (server-side problem). Try again later.
      </code>
    </td>
    
    <td>
      The server-side audio probe failed
    </td>
    
    <td>
      Retry later, and tell an administrator
    </td>
  </tr>
  
  <tr>
    <td>
      <code>
        The upstream service timed out. Try again later
      </code>
    </td>
    
    <td>
      The speech recognition service took too long to respond
    </td>
    
    <td>
      Retry later. Longer audio times out more often
    </td>
  </tr>
  
  <tr>
    <td>
      <code>
        The service restarted. Trigger it again
      </code>
    </td>
    
    <td>
      The service restarted mid-processing
    </td>
    
    <td>
      Trigger a batch task again; retry a single transcription
    </td>
  </tr>
</tbody>
</table>

### Which statuses can be retried or canceled

Retry and cancel each apply to particular statuses; outside those, the action icon doesn't appear:

<table>
<thead>
  <tr>
    <th>
      Action
    </th>
    
    <th>
      Applies to
    </th>
    
    <th>
      Message when it doesn't
    </th>
  </tr>
</thead>

<tbody>
  <tr>
    <td>
      Retry
    </td>
    
    <td>
      <strong>
        Failed
      </strong>
      
      , <strong>
        Canceled
      </strong>
    </td>
    
    <td>
      <code>
        This transcription task can't be retried
      </code>
    </td>
  </tr>
  
  <tr>
    <td>
      Cancel
    </td>
    
    <td>
      <strong>
        Queued
      </strong>
      
      , <strong>
        Processing
      </strong>
    </td>
    
    <td>
      <code>
        This transcription task can't be canceled
      </code>
    </td>
  </tr>
</tbody>
</table>

**Skipped** can't be retried. That status means a successful transcription already exists, so retrying is pointless — see [Why are some files marked as skipped?](https://docs.ardge.com/en/knowledge/scribestt/skipped-files)

## How to retry

To retry a single transcription:

1. Go to the **Transcriptions** page
2. In that record's **Status** column, click the retry icon next to the status label

The page shows "Transcription task resubmitted", the record's **Status** returns to **Queued**, and it moves on to **Processing**.

<warning>

**A retry doesn't update its run's counts:**

A batch run's **Total**, **Succeeded**, **Failed**, and **Skipped** are a snapshot from scan time. Retrying one transcription successfully doesn't change the counts in **Run history** — go by that file's **Status** in **Run details** instead.

</warning>

## Related articles

- [Set up the speech recognition model provider in ScribeSTT](https://docs.ardge.com/en/knowledge/scribestt/model-provider)
- [What file formats and size limits does ScribeSTT support?](https://docs.ardge.com/en/knowledge/scribestt/formats-and-limits)
- [Why are some files marked as skipped?](https://docs.ardge.com/en/knowledge/scribestt/skipped-files)


## Sitemap

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