@Alberto Fernandez Martinez I think the error indicates one or all of your environment variables did not get correctly populated. Could you try to use them directly in your script for testing purpose? I have used the following script with my whisper model deployment and it seems to work as expected.
Also, ensure you have deployed your whisper base model as a deployment through Azure OpenAI studio before using it in the client call to create transcription.
import os
from openai import AzureOpenAI
from pathlib import Path
client = AzureOpenAI(api_key="your_key", api_version="2023-10-01-preview", azure_endpoint ="https://your_resource.openai.azure.com")
client.audio.transcriptions.create(model = "your_whisper_deployment_name", file = Path("path_to_file"), response_format = "text" )