This browser is no longer supported.
Upgrade to Microsoft Edge to take advantage of the latest features, security updates, and technical support.
![](data:image/svg+xml, %3Csvg xmlns='http://www.w3.org/2000/svg' height='64' class='font-weight-bold' style='font: 600 30.11764705882353px "SegoeUI", Arial' width='64'%3E%3Ccircle fill='hsl(124.80000000000001, 50%, 22%)' cx='32' cy='32' r='32' /%3E%3Ctext x='50%25' y='55%25' dominant-baseline='middle' text-anchor='middle' fill='%23FFF' %3EV%3C/text%3E%3C/svg%3E)
I am trying to create and test a custom speech model. I'm able to go through all of the steps to upload data, train the model, and test the model. However, I can't view the contents of files after I upload them (for example, a plain text file has a large gray box where it should display the text the file contains) or the transcripts produced by the model as part of a text (again, a gray box, though I can listen to the audio file it should have transcribed). Any attempt to download the file or test results to view the contents will fail. The upload of the file and the running of the test are listed as being successful, it's just my ability to view the results that isn't working.
I'm guessing this is coming from a permissions error somewhere. I am a Cognitive Services Speech Contributor for this resource, which should give me full access for reading data for custom speech models. Is the training data stored somewhere else, so I need to be a Cognitive Services Speech Contributor for that location as well? If so, where can I find the storage location? Or does the role Cognitive Services Speech Contributor not contain the permissions I need to read the data I upload? If so, what role would allow me to do this?
@Victoria Are you referring to the Test Models screen from the speech studio where the download option is available?
If you are referring to a different screen, could you please add the screen shot?
Based on what is documented I think you would need a role to list keys. Cognitive Services Speech Contributor does not have access to list keys of cognitive services resource. Please see this page for details.
I would recommend to add Cognitive Services Contributor role to this user and check if the required action is enabled. Please ensure to logout and login to check if the changes take effect. Thanks!!
I accidentally sent my response as an answer to the overall question as opposed to a reply to your comment. If you check the answer, you can see the screenshots of the two screens I'm talking about.
Did you try assigning the Cognitive Services Contributor role to check if it loads the page successfully?
I already had the Cognitive Services Contributor role.
The solution ended up being to enable Resource Sharing (CORS) on the BYOS storage account connected to the project.
It required an origin of https://speech.microsoft.com , GET and OPTIONS for allowed methods, * for allowed and exposed headers, and 1000 for the max age.
I'm talking about two screens. When I say I can't view the contents of an uploaded file, I mean the Speech Datasets page, when I click to view more of a successfully uploaded plain text file:
When I say I can't view the results of a test, I do mean the Test Models page, but specifically when I click on one test to view the custom-model-created transcripts for the audio files used in that test.
