MiniCPM-o-4_5 transformer推理"video_url": {"url": video_path, "use_audio": True} use_audio默认都为True不可更改么？

[KivenJoo]

在接入MiniCPM-o-4_5时，视频推理部分，采用OpenAI的结构推理，content.append({
"type": "video_url",
"video_url": {"url": video_path, "use_audio": False}
}) 使用"use_audio": False 发现未生效，我的视频如果没有音频的时候会报错

answer = self.model.chat(msgs=messages, max_new_tokens=32768, omni_mode=True, use_tts_template=False,
File "/usr/local/conda/lib/python3.10/site-packages/torch/utils/_contextlib.py", line 116, in decorate_context
return func(*args, **kwargs)
File "/home/hadoop-aipnlp/.cache/huggingface/modules/transformers_modules/main/modeling_minicpmo.py", line 1146, in chat
content = normalize_content(content)
File "/home/hadoop-aipnlp/.cache/huggingface/modules/transformers_modules/main/utils.py", line 2400, in normalize_content
normalized = normalize_content_item(item)
File "/home/hadoop-aipnlp/.cache/huggingface/modules/transformers_modules/main/utils.py", line 2338, in normalize_content_item
video_frames, audio_segments, stacked_frames = get_video_frame_audio_segments(
File "/home/hadoop-aipnlp/.local/lib/python3.10/site-packages/minicpmo/utils.py", line 414, in get_video_frame_audio_segments
audio_segments = get_audio_segments(
File "/home/hadoop-aipnlp/.local/lib/python3.10/site-packages/minicpmo/utils.py", line 175, in get_audio_segments
video_clip.audio.write_audiofile(temp_audio_file_path, codec="pcm_s16le", fps=sr)
AttributeError: 'NoneType' object has no attribute 'write_audiofile'

看了下https://huggingface.co/openbmb/MiniCPM-o-4_5/blob/main/utils.py#L2338 好像没有支持use_audio 这个参数的传递？这个目前有其他方法解决么？

[airlsyn]

@KivenJoo We've fixed it. Please update the following:

1. Re-download utils.py from openbmb/MiniCPM-o-4_5
2. Upgrade minicpmo-utils to version >=1.0.6:
   pip install --upgrade "minicpmo-utils>=1.0.6"

After that, "use_audio": False should work as expected with your current input. Feel free to follow up if you run into any other issues.