loading audio data from an array

moss_transcribe_diarize\inference_utils.py
def process_audio_info
Judging by the code, the ability to load audio data from an array was intended, but it doesn't work due to an error
audio = item.get("audio") or item.get("audio_url") or item.get("url") or item.get("path")
ValueError: The truth value of an array with more than one element is ambiguous. Use a.any() or a.all()

You can fix it like this, for example:
#audio = item.get("audio") or item.get("audio_url") or item.get("url") or item.get("path")
audio = None
for key in ["audio", "audio_url", "url", "path"]:
if item.get(key) is not None:
audio = item[key]
break

It would also be worth fixing build_transcription_messages or making a separate function for loading from an array

Start in moss_transcribe_diarize/inference_utils.py at process_audio_info and inspect how audio, audio_url, url, and path are selected. Verify that array-valued audio input no longer raises the ambiguous truth-value error, while the other supported input forms still work; also review build_transcription_messages as suggested for related array handling.