Azure Speech Studio taking way too long to train custom model and when does the pricing start?

Question

I have a 20 KB plain text file with a maximum of 300 lines that I want to use for training. The format is correct, and I’m using the Default Baseline Model 20240327. However, the training process seems stuck—it’s been running for 2-3 hours with no progress updates, and there’s nothing reflected in the cost report.

I can open Speech Studio, create a dataset, and put a model to train, so I should have all the permissions right?

For context, I have the following roles assigned to the resource group:

• Cognitive Services User
• Cognitive Services Speech Contributor
• Cognitive Services Speech User

Here are my questions:

1. How long should it typically take to fine-tune with a 20 KB dataset?
2. Am I incurring any costs during preprocessing steps, such as model loading or network delays?
3. Could I be missing something, like additional roles or access permissions?