fix(python/voice-live-avatar): replace retired gpt-4o-realtime, align with JS sample

[zhifzhan]

Summary

Fix Python voice-live-avatar demo failing to connect with GPT Realtime (model id was retired) and align its model + voice options with the JavaScript sample.

Symptoms

Clicking Connect with the default GPT Realtime selection produced Error: SESSION_UPDATED event not received. The server log showed Azure rejecting the request with:

Model 'gpt-4o-realtime' has been retired. Please use 'gpt-realtime' instead.

Changes

• Replace retired model ids gpt-4o-realtime / gpt-4o-mini-realtime with gpt-realtime / gpt-realtime-mini across static/index.html, app.py, voice_handler.py, README.md.
• Extend the Model dropdown to match the JS sample: add gpt-realtime-1.5, azure-realtime, full gpt-5* cascaded family; drop phi4-mm and phi4-mini.
• Update the cascaded-model list in static/app.js so cascaded-only UI fields toggle correctly.
• Add azure-realtime-native voice handling, mirroring the JS sample:
  • New voice type option (visible only when model is azure-realtime).
  • New native voice selector with 11 voices (Aarti, Andrew, Ava, Denise, Elsa, Florian, Francisca, Meera, Xiaoxiao, Ximena, Yunxi).
  • Selecting azure-realtime auto-switches voice type to native and defaults voice to ava; switching away resets to standard.
  • Voice speed field is hidden for native voices.
• voice_handler._build_voice_config constructs the azure-realtime-native voice via AzureVoice({"type": ..., "name": ...}) since the SDK has no typed subclass for this voice type yet.

Notes

Verified end-to-end against an Azure AI Services endpoint: all 19 models in the new dropdown reach session_started successfully, including azure-realtime with the native voice override.

[Copilot]

Pull request overview

This PR updates the Python voice-live-avatar sample to use the current GPT Realtime model IDs and aligns the frontend model/voice options with the JavaScript sample, including support for azure-realtime native voices.

Changes:

• Replaced retired gpt-4o-*-realtime defaults with gpt-realtime* in the Python backend and docs.
• Expanded the model dropdown and updated the cascaded-model list to keep cascaded-only UI fields toggling correctly.
• Added azure-realtime-native voice support (new UI selector + backend voice config construction).

Reviewed changes

Copilot reviewed 5 out of 5 changed files in this pull request and generated 3 comments.

Show a summary per file

File	Description
python/voice-live-avatar/voice_handler.py	Updates default model and adds backend construction for azure-realtime-native voice config.
python/voice-live-avatar/static/index.html	Updates model dropdown options and adds UI elements for selecting native voices.
python/voice-live-avatar/static/app.js	Adds model-change behavior for azure-realtime, updates cascaded model list, and toggles native-voice UI fields.
python/voice-live-avatar/README.md	Updates documented default model value.
python/voice-live-avatar/app.py	Updates backend-provided default model returned to the frontend.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.