I am experiencing an issue with the Document Translation API (api-version: 2025-12-01-preview) when translating PNG image files.

Question

I am experiencing an issue with the Document Translation API (api-version: 2025-12-01-preview) when translating PNG image files.

邱凌东 0

API Endpoint:

POST /translator/document/batches

Issue Description:

PNG image translation fails consistently with the following error

{
    "id": "0ef5ccf0-2746-45e8-b712-24b165eeb675",
    "createdDateTimeUtc": "2026-03-10T10:20:56.3105795Z",
    "lastActionDateTimeUtc": "2026-03-10T10:21:07.7822589Z",
    "status": "Failed",
    "error": {
        "code": "InternalServerError",
        "message": "Document failed during parsing. If the error persists, report it with date/time of error, request identifier from response header X-RequestId, and client identifier from request header X-ClientTraceId.",
        "target": "Operation",
        "innerError": {
            "code": "InternalServerErrorDocumentParsing",
            "message": "Document failed during parsing. If the error persists, report it with date/time of error, request identifier from response header X-RequestId, and client identifier from request header X-ClientTraceId."
        }
    },
    "summary": {
        "total": 1,
        "failed": 1,
        "success": 0,
        "inProgress": 0,
        "notYetStarted": 0,
        "cancelled": 0,
        "totalCharacterCharged": 0,
        "totalImageCharged": 0
    }
}

邱凌东 0 Reputation points

2026-03-10T10:29:30.7+00:00

This is how images are translated from Chinese to Korean.
Anshika Varshney 10,145 Reputation points Microsoft External Staff Moderator

2026-04-20T04:52:48.32+00:00

Hi 邱凌东.

Thanks for sharing the detailed error, this is helpful.

From the error message and behavior, it looks like the issue is happening during document parsing on the service side, not because of your request format or configuration.

The important part in your error is that the status is failed with message Document failed during parsing and inner error also points to parsing failure.

Also, I noticed that no characters or images were processed in the summary. This usually means the service was not able to read or process the file at all, so the failure happened very early.

Here are a few things you can check from your side:

First, validate the PNG file itself Try opening the file locally and make sure it is a valid image and not corrupted, If possible, try with another PNG file to see if the issue is specific to that file.

Second, check the image format and size Even though PNG is supported in preview, some edge cases like unusual encoding, very large file size, or unsupported image properties can cause parsing issues.

Third, try converting the same PNG to another format like JPEG or PDF and test again If that works, it may confirm that the issue is related to how that specific PNG is encoded.

Fourth, try a very simple test file Use a small PNG with clear text and run the same API call This helps confirm whether the issue is general or only for specific files.

Based on the error, this does not look like a problem with endpoint, authentication, or request structure It is more likely related to how the service is handling that particular image in this preview version.

Also keep in mind that this feature is currently in preview, so some formats or files may not work consistently yet.

In short: your request looks correct the file is reaching the service but the service is not able to parse that image successfully.

Trying with a different or simplified image should help confirm the root cause.

I Hope this helps. Do let me know if you have any further queries.

Thankyou!
SRILAKSHMI C 18,035 Reputation points Microsoft External Staff Moderator

2026-04-22T04:50:11.67+00:00

Hi @邱凌东

We haven’t heard from you on the last response and was just checking back to see if you have a resolution yet. In case if you have any resolution please do share that same with the community as it can be helpful to others. Otherwise, will respond with more details and we will try to help.

Thank you!
Manas Mohanty 16,670 Reputation points Microsoft External Staff Moderator

2026-04-29T03:02:48.56+00:00
邱凌东

Good day. Manas here working on this case.

Hope you found the inputs on image requirements useful

it seems you are facing internal server error issue while translating documents from Korean to English.

Which points towards supportability of that image (high graphics content here) though png is one supported format within preview nature of this image translation feature. All regions might not be working (Normally East US /West Europe and other limited regions.

Replicating on my side to be on same page.

Other options to consider for testing

vision-based models from OpenAI along AI search indexed with OCR/other custom skills

Proposed approach for better accuracy - Do OCR using prebuilt model+ translate using text translate API

Follow up queries

Please share the region and sample code used at your side

Could you confirm the presence of VNET in private message

Reference docs used

https://learn.microsoft.com/en-us/azure/ai-services/translator/document-translation/reference/start-batch-translation#translate-image-files

https://techcommunity.microsoft.com/blog/azure-ai-foundry-blog/translate-documents-using-azure-document-translation/3796919

Thank you

2 answers

Your answer

邱凌东 0 Reputation points

2026-03-10T10:29:30.7+00:00

This is how images are translated from Chinese to Korean.
SRILAKSHMI C 18,035 Reputation points Microsoft External Staff Moderator

2026-04-22T04:50:11.67+00:00

Hi @邱凌东

We haven’t heard from you on the last response and was just checking back to see if you have a resolution yet. In case if you have any resolution please do share that same with the community as it can be helpful to others. Otherwise, will respond with more details and we will try to help.

Thank you!
Manas Mohanty 16,670 Reputation points Microsoft External Staff Moderator

2026-04-29T03:02:48.56+00:00

邱凌东

Good day. Manas here working on this case.

Hope you found the inputs on image requirements useful

it seems you are facing internal server error issue while translating documents from Korean to English.

Which points towards supportability of that image (high graphics content here) though png is one supported format within preview nature of this image translation feature. All regions might not be working (Normally East US /West Europe and other limited regions.

Replicating on my side to be on same page.

Other options to consider for testing

vision-based models from OpenAI along AI search indexed with OCR/other custom skills

Proposed approach for better accuracy - Do OCR using prebuilt model+ translate using text translate API

Follow up queries

Please share the region and sample code used at your side

Could you confirm the presence of VNET in private message

Reference docs used

https://learn.microsoft.com/en-us/azure/ai-services/translator/document-translation/reference/start-batch-translation#translate-image-files

https://techcommunity.microsoft.com/blog/azure-ai-foundry-blog/translate-documents-using-azure-document-translation/3796919

Thank you

Answer 1

Hi 邱凌东

I just tested below batch translation code in West US 2 region and getting 202 response code instead.

Could you test in West US 2 with below code and let us know your observation.

import requests
import json

# ========= CONFIG =========
TRANSLATOR_KEY = "<apikey>"
TRANSLATOR_ENDPOINT = "https://<translation>.cognitiveservices.azure.com"

API_VERSION = "2025-12-01-preview"

# ✅ SINGLE FILE (not container)

SOURCE_FILE_URL = <SAS URL of file with READ, write, create permission>
TARGET_CONTAINER_URL = <SAS Url of target container Read, write, create>

SOURCE_LANGUAGE = "zh-Hans"
TARGET_LANGUAGE = "ko"

# ========= REQUEST =========
url = f"{TRANSLATOR_ENDPOINT}/translator/document/batches?api-version={API_VERSION}"

headers = {
    "Ocp-Apim-Subscription-Key": TRANSLATOR_KEY,
    "Content-Type": "application/json"
}

payload = {
    "inputs": [
        {
            "storageType": "File",   # ✅ REQUIRED for single file
            "source": {
                "sourceUrl": SOURCE_FILE_URL,
                "language": SOURCE_LANGUAGE,
                
            },
            "targets": [
                {
                    "targetUrl": TARGET_CONTAINER_URL,
                    "language": TARGET_LANGUAGE,
                   
                }
            ]
        }
    ]
}

response = requests.post(url, headers=headers, data=json.dumps(payload))

# ========= RESPONSE =========
print("Status Code:", response.status_code)

if response.status_code == 202:
    print("Batch submitted successfully.")
    print("Operation-Location:", response.headers.get("Operation-Location"))
else:
    print("Error:")
    print(response.text)

Thank you.

Answer 2

PNG image translation with the Document Translation API using api-version=2025-12-01-preview is supported as part of the image translation preview feature.

The error payload shows:

code: InternalServerError
innerError.code: InternalServerErrorDocumentParsing
message: "Document failed during parsing. If the error persists, report it with date/time of error, request identifier from response header X-RequestId, and client identifier from request header X-ClientTraceId."

This indicates a service-side parsing failure rather than a validation or configuration problem. For image translation with this preview API version, no additional configuration is required beyond submitting a supported image file type (such as .png) in a standard batch request.

Given that:

.png is explicitly listed as a supported image format for this preview.
The error is an internal parsing error with zero characters/images charged.

The next step is to follow the guidance in the error message and report the issue to Microsoft with:

Date and time of the failure.
The X-RequestId value from the HTTP response headers.
The X-ClientTraceId value from the request headers (if set).
The failing PNG sample (or a minimal repro file) and the exact api-version and endpoint used.

This will allow the product team to investigate the parsing failure in the preview image translation pipeline.

References:

Share via

I am experiencing an issue with the Document Translation API (api-version: 2025-12-01-preview) when translating PNG image files.

2 answers

Your answer