You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I have a CutSet and I want to define a custom Resampler, for example a neural model that performs bandwidth extension from 16 kHz to 24 kHz and then extract features.
Or in general create a transformation that will be applied to the CutSet in a similar way to .resample, .perturb_volume etc.
I found that there is no easy way to do this, there is no documentation as to how you apply transforms to the recordings apart from the predefined ones.
The text was updated successfully, but these errors were encountered:
That's correct, currently adding a new recording-level transform is a little complicated. This PR shows all necessary steps by adding volume perturbation: #382
I'll try to add some docs how to add new transforms and expose a new API to make it simpler to apply them.
I have a CutSet and I want to define a custom Resampler, for example a neural model that performs bandwidth extension from 16 kHz to 24 kHz and then extract features.
Or in general create a transformation that will be applied to the CutSet in a similar way to .resample, .perturb_volume etc.
I found that there is no easy way to do this, there is no documentation as to how you apply transforms to the recordings apart from the predefined ones.
The text was updated successfully, but these errors were encountered: