Add Compose Transform about audiomentations

[iver56]

Can you give an example? How is this different from the existing Compose?

[zh794390558]

We can do two Transform controlled by one Prob.

[iver56]

We can make a callable that applies two transforms with a single probability like this:

Compose([AddGaussianNoise(p=1.0), Normalize(p=1.0)], p=0.5)

Does this satisfy your requirement, or would you like to elaborate?

[zh794390558]

But how Compose([Compose([AddGaussianNoise(p=1.0), Normalize(p=1.0)], p=0.5), ...]) this work?

[iver56]

So you want to nest Compose instances, as in the following code example?

Compose(
    [
        Compose([AddGaussianNoise(p=1.0), Normalize(p=1.0)], p=0.5),
        Compose([AddBackgroundNoise(p=1.0, ...), Normalize(p=1.0)], p=0.5),
    ]
)

Can you tell me more about your underlying use case?

[zh794390558]

We can using AddShortNoise add some noises one time.

[iver56]

And how does that relate to shortcomings in Compose? I'll reopen this issue if it gets a proper description that I can understand.