Add vLLM version compatibility for AudioMediaIO changes

[Copilot]

vLLM's AudioMediaIO class was removed or relocated in recent versions, breaking import with AttributeError: module 'vllm.multimodal.audio' has no attribute 'AudioMediaIO'.

Changes

• Version-agnostic AudioMediaIO handling: Try-except pattern detects presence of AudioMediaIO at import time
  • Old vLLM: inherit from AudioMediaIO and override methods with FFmpeg implementations
  • New vLLM: define standalone class with identical interface
• Diagnostic warnings: Log which compatibility path was taken to aid debugging
• Broader exception coverage: Catch both AttributeError and ImportError for robustness

try:
    _OriginalAudioMediaIO = _vllm_audio_module.AudioMediaIO
    class _PatchedAudioMediaIO(_OriginalAudioMediaIO):
        # Override with FFmpeg implementations
except (AttributeError, ImportError) as e:
    warnings.warn(f"AudioMediaIO not found ({e}). Using standalone implementation.")
    class _PatchedAudioMediaIO:
        # Standalone FFmpeg-based implementation

FFmpeg-based audio loading behavior remains unchanged across all vLLM versions.

Warning

Firewall rules blocked me from connecting to one or more addresses (expand for details)

I tried to connect to the following addresses, but was blocked by firewall rules:

• docs.vllm.ai
  • Triggering command: /home/REDACTED/work/_temp/ghcca-node/node/bin/node /home/REDACTED/work/_temp/ghcca-node/node/bin/node --enable-source-maps /home/REDACTED/work/_temp/copilot-developer-action-main/dist/index.js (dns block)

If you need me to access, download, or install something from one of these locations, you can either:

• Configure Actions setup steps to set up my environment, which run before the firewall is enabled
• Add the appropriate URLs or hosts to the custom allowlist in this repository's Copilot coding agent settings (admins only)

Original prompt

---

This section details on the original issue you should resolve

<issue_title>vllm asr problem</issue_title>
<issue_description>I am currently following the document VibeVoice vLLM ASR Deployment
to employ the vibevoice model. But I encountered this problem:

Traceback (most recent call last):
  File "<frozen runpy>", line 189, in _run_module_as_main
  File "<frozen runpy>", line 112, in _get_module_details
  File "/app/vllm_plugin/__init__.py", line 17, in <module>
    from .model import VibeVoiceForCausalLM
  File "/app/vllm_plugin/model.py", line 76, in <module>
    _OriginalAudioMediaIO = _vllm_audio_module.AudioMediaIO
                            ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
AttributeError: module 'vllm.multimodal.audio' has no attribute 'AudioMediaIO'
Traceback (most recent call last):
  File "/app/vllm_plugin/scripts/start_server.py", line 170, in <module>

============================================================
  VibeVoice vLLM ASR Server - One-Click Deployment
    main()
  File "/app/vllm_plugin/scripts/start_server.py", line 163, in main
============================================================

============================================================
  Updating package list
============================================================


============================================================
  Installing FFmpeg and audio libraries
============================================================


============================================================
    generate_tokenizer(model_path)
  Installing VibeVoice with vLLM support
============================================================


============================================================
  Generating tokenizer files
============================================================

  File "/app/vllm_plugin/scripts/start_server.py", line 73, in generate_tokenizer
    run_command(
  File "/app/vllm_plugin/scripts/start_server.py", line 30, in run_command
    subprocess.run(cmd, check=True)
  File "/usr/lib/python3.12/subprocess.py", line 571, in run
    raise CalledProcessError(retcode, process.args,
subprocess.CalledProcessError: Command '['/usr/bin/python3', '-m', 'vllm_plugin.tools.generate_tokenizer_files', '--output', '/models/VibeVoice-ASR']' returned non-zero exit status 1.

The docker command I use is:

docker run -d --gpus '"device=2"' --name vibevoice-vllm \
  --ipc=host \
  -p 8000:8000 \
  -e VIBEVOICE_FFMPEG_MAX_CONCURRENCY=64 \
  -e PYTORCH_ALLOC_CONF=expandable_segments:True \
  -e HF_ENDPOINT=https://hf-mirror.com \
  -e VLLM_USE_V1=0 \
  -v $(pwd):/app \
  -v /141nfs/xxx/hf_models:/models \
  -w /app \
  --entrypoint bash \
  swr.cn-north-4.myhuaweicloud.com/ddn-k8s/docker.io/vllm/vllm-openai:latest \
  -c "python3 /app/vllm_plugin/scripts/start_server.py --model /models/VibeVoice-ASR"

Note that to save the time of downloading the model, I already download the VibeVoice-ASR on my local device and changed the corresponding loading procedure:

    # Step 3: Download model
    # model_path = download_model(args.model)
    model_path = args.model
```</issue_description>

## Comments on the Issue (you are @copilot in this section)

<comments>
</comments>

• Fixes vllm asr problem #222

---

✨ Let Copilot coding agent set things up for you — coding agent works faster and does higher quality work when set up for your repo.