OutputKey in start_transcription_job() cant have `=`

[sidatcd]

Describe the bug

An error occurs when calling the start_transcription_job() operation in AWS Transcribe with an OutputKey containing an equals sign (=).
This leads to a BadRequestException, even though S3 allows keys with "=".

Error

An error occurred (BadRequestException) when calling the StartTranscriptionJob operation: 1 validation error detected: Value 'name=name/stage=prod/2b32d9cd6ce13621d332e1ae23d7a7ba.json' at 'outputKey' failed to satisfy constraint: Member must satisfy regular expression pattern: [a-zA-Z0-9-_.!*'()/]{1,1024}$ 

Regression Issue

• Select this option if this issue appears to be a regression.

Expected Behavior

The OutputKey should allow the equals sign ("=") as it's a valid character in S3 object keys.

Current Behavior

Throws the regex validation error

Reproduction Steps

1. Call start_transcription_job() with an OutputKey that includes "=" (e.g., name=name/stage=prod/...).
2. Observe the error.

Possible Solution

No response

Additional Information/Context

No response

SDK version used

boto3==1.35.44

Environment details (OS name and version, etc.)

mac

[sidatcd]

API has the same behaviour.

[adev-code]

Hello @sidatcd, thanks for reaching out. For a further look, please provide code snippet and full debug logs by adding the line of code: boto3.set_stream_logger('') redacting any sensitive information. Thank you.

[sidatcd]

Hello @sidatcd, thanks for reaching out. For a further look, please provide code snippet and full debug logs by adding the line of code: boto3.set_stream_logger('') redacting any sensitive information. Thank you.

Hello @adev-code
From my further investigation, this is an issue with StartTranscriptionJob API itself.Default validation does not allow =.
Same error when implemented in step function or other sdks, manually copying after setting transcribe output to a temp bucket as a workaround now.