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
To be able to replicate results from some models, e.g. Sainburg et al 2020 and Goffinet 2021, it would be good to use the exact functions they use to compute spectrogram. This could also be helpful for users more generally.
We should achieve this by allowing a user to specify the callable / function to use to make spectrograms.
I imagine in a config this would look something like
Where spect_params can be any kwargs that are passed into the Callable specified by spect_maker
Under the hood we will need to do
make sure we can import the callable
document what makes it possible to do so, i.e., have it installed or have it in the current working directory
determine dynamically if it's a class we need to instantiate -- can we do this? I think yes, with isclass
pass spect_params into the spect_maker
We should probably do this at the same time we switch to using vocalpy as a backend in the prep step as in #558
The text was updated successfully, but these errors were encountered:
NickleDave
changed the title
ENH: Add ability to specify spectrogram function with vocalpy to prep step
ENH: Add ability to specify function/callable to make spectrogram during prep step
Jul 18, 2023
To be able to replicate results from some models, e.g. Sainburg et al 2020 and Goffinet 2021, it would be good to use the exact functions they use to compute spectrogram. This could also be helpful for users more generally.
We should achieve this by allowing a user to specify the callable / function to use to make spectrograms.
I imagine in a config this would look something like
Where
spect_params
can be anykwargs
that are passed into theCallable
specified byspect_maker
Under the hood we will need to do
isclass
spect_params
into thespect_maker
We should probably do this at the same time we switch to using
vocalpy
as a backend in the prep step as in #558The text was updated successfully, but these errors were encountered: