Unable to use GCP signed URL with Azure Cognitive Speech to text

Question

I have uploaded an audio file on GCS bucket and then I created a signed URL for the file using this method:

storage.signUrl(
                blobInfo,
                DURATION,
                TimeUnit.DAYS,
                Storage.SignUrlOption.httpMethod(HttpMethod.GET),
                Storage.SignUrlOption.withContentType()
        );

When I pass this URL to Azure cognitive Speech to text service using the batch transcription REST API this is the request body of azure's create transcription API

{
  "contentUrls": [
    "gcpsignedurl"
  ],
  "properties": {
    "diarizationEnabled": false,
    "wordLevelTimestampsEnabled": false,
    "punctuationMode": "DictatedAndAutomatic",
    "profanityFilterMode": "Masked"
  },
  "locale": "en-US",
  "displayName": "Transcription using default model for en-US"
}

the transcription job gets created and I get the location URL of transcription from the header, now when I Am calling the location URL I get this error

    "error": {
      "code": "InvalidUri",
      "message": "Authentication failed for recordings URI."
    }
    "status": "Failed"

how do I solve this problem?

Answer 1

I think you are using Speech to Text API V3.0 - Batch transcription, it can read audio from a public-visible internet URI , and can read audio or write transcriptions using a SAS URI with Azure Blob storage. You need to make sure your GCP url is public and can be read without any authentication.