skip audio loading if no audio in video

[Artiprocher]

No description provided.

[gemini-code-assist]

Code Review

This pull request updates the version to 2.0.9 and adds a try-except block to handle audio loading failures in operators.py. The review identified critical issues in the implementation: returning a single None value will cause unpacking errors in callers, the bare except block should be more specific, and the imageio reader needs proper resource management via a context manager.

[gemini-code-assist]

The current implementation has several issues:

1. Unpacking Error: The method is expected to return a tuple (waveform, sample_rate). Returning None in the except block will cause a TypeError when the caller attempts to unpack the result (e.g., waveform, sr = loader(data)). Returning None, None ensures the caller can still unpack the result.
2. Bare except: block: Using except: without specifying an exception type catches all exceptions, including SystemExit and KeyboardInterrupt. It is better to catch Exception or specific errors like RuntimeError from torchaudio.
3. Resource Leak: The imageio reader is opened but never closed. Using a with statement ensures the reader is properly closed even if an exception occurs during frame counting.

Suggested change

	try:
	reader = self.get_reader(data)
	num_frames = self.get_num_frames(reader)
	duration = num_frames / self.frame_rate
	waveform, sample_rate = torchaudio.load(data)
	target_samples = int(duration * sample_rate)
	current_samples = waveform.shape[-1]
	if current_samples > target_samples:
	waveform = waveform[..., :target_samples]
	elif current_samples < target_samples:
	padding = target_samples - current_samples
	waveform = torch.nn.functional.pad(waveform, (0, padding))
	return waveform, sample_rate
	except:
	warnings.warn(f"Cannot load audio in {data}. The audio will be `None`.")
	return None
	try:
	with self.get_reader(data) as reader:
	num_frames = self.get_num_frames(reader)
	duration = num_frames / self.frame_rate
	waveform, sample_rate = torchaudio.load(data)
	target_samples = int(duration * sample_rate)
	current_samples = waveform.shape[-1]
	if current_samples > target_samples:
	waveform = waveform[..., :target_samples]
	elif current_samples < target_samples:
	padding = target_samples - current_samples
	waveform = torch.nn.functional.pad(waveform, (0, padding))
	return waveform, sample_rate
	except Exception:
	warnings.warn(f"Cannot load audio in {data}. The audio will be `None`.")
	return None, None