Fix Python FastAPI middle tier demo for Azure backend

[kingychiu]

Purpose

1. The original code force people to use Azure CLI to login, this PR add another option (using the API Key generated from Azure Portal)

2. The client.send_item expects the message to have an id attribute because it expects an Item type, so the current demo will raise an exception when user try to send a text message. Updating to UserMessageItem type has default id=None attribute.

3. Change the default voice to alloy, because coral voice also trigger a runtime exception.

4. The import order changes are from ruff linter & format, if you wish I can revert those.

Does this introduce a breaking change?

[ ] Yes
[X] No

Pull Request Type

What kind of change does this Pull Request introduce?

[X] Bugfix
[ ] Feature
[ ] Code style update (formatting, local variables)
[ ] Refactoring (no functional changes, no api changes)
[ ] Documentation content changes
[ ] Other... Please describe:

How to Test

• Setup and start the backend under: samples/middle-tier/python-fastapi
• Setup and start the frontend under: samples/middle-tier/generic-frontend
• Send a text message for a response.

Other Information

I found another issue (not fixed by this PR): the RTClient.generate_response conflicts with RTClient.events, because both will take the response.created event. As a result

• RTClient.generate_response will be blocked forever, because response.created is removed by RTClient.events
• In the demo here, we cannot send text messages more than once, because the first text message will block the process.

I am creating another issue for this: #111

[kingychiu]

@kingychiu please read the following Contributor License Agreement(CLA). If you agree with the CLA, please reply with the following information.

@microsoft-github-policy-service agree [company="{your company}"]

Options:

• (default - no company specified) I have sole ownership of intellectual property rights to my Submissions and I am not making Submissions in the course of work for my employer.

@microsoft-github-policy-service agree

• (when company given) I am making Submissions in the course of work for my employer (or my employer has intellectual property rights in my Submissions by contract or applicable law). I have permission from my employer to make Submissions and enter into this Agreement on behalf of my employer. By signing below, the defined term “You” includes me and my employer.

@microsoft-github-policy-service agree company="Microsoft"

Contributor License Agreement

@microsoft-github-policy-service agree

[ElliotRoe]

Please merge 🙏 I am also have these troubles with the Python middle tier

[ElliotRoe]

@kingychiu are you maintaining an updated version of the rtclient package? I'd be happy to contribute if so.

[kingychiu]

@ElliotRoe, If you clone the code from this PR, it should work except the text input issue I mentioned here #111.

Because I need both audio and text input, I switched to using OpenAI-python beta realtime api.
https://github.com/openai/openai-python/tree/main?tab=readme-ov-file#realtime-api-beta

Init client:

client = AsyncAzureOpenAI(
        azure_endpoint="AZURE_OPENAI_ENDPOINT",
        api_key="AZURE_OPENAI_KEY",
        api_version="2024-10-01-preview",
    )

Connection:

async with self.client.beta.realtime.connect(
    model="gpt-4o-realtime-preview"
) as conn:

The rest of the websocket logic is mostly similar to this aoai-realtime-audio-sdk repo, except I didn't implement the locking I mentioned in the text input issue linked above.