Closed Rahul-Brito closed 5 months ago
i thought i checked two types of files. but this is where tensor shapes on different channel data can be different. if you simply load it what does the shape look like?
i would change that to:
if len(signal.shape) > 1 and ...
...
else:
signal = signal[:, None]
That audio looks like this:
It was a mono 44khz audio I recorded in audacity.
I can make the change
https://github.com/sensein/b2aiprep/blob/ac73e7b1cf043415d2bec1a2338755c1199b9ce6/src/b2aiprep/process.py#L34
When I run from_file on data/test_audio.wav I get the error:
IndexError: too many indices for tensor of dimension 1
Since the signal that gets loaded only has one dimension. Is this a data issue or is the code handling the situation incorrectly?