vak.transforms.transforms.AddChannel#

class vak.transforms.transforms.AddChannel(channel_dim: int | float = 0)[source]#

Bases: object

Add a “channel” dimension to a tensor.

Transform that makes it easy to treat a spectrogram as an image, by adding a dimension with a single ‘channel’, analogous to grayscale. In this way the tensor can be fed to e.g. convolutional layers.

Parameters:

  • input (torch.Tensor) – with two dimensions (height, width).

  • channel_dim (int) – dimension where “channel” is added. Default is 0, which returns a tensor with dimensions (channel, height, width).

__init__(channel_dim: int | float = 0)[source]#

Methods

__init__([channel_dim])