Brian McFee
Brian McFee
My $2e-2: i think it makes sense to have a self-contained `bsseval` package and either deprecate the implementation in `mir_eval`, or refactor it to depend on the `bsseval` implementation. A...
tagging @ejhumphrey because of course
> I think an absolute pitch class would will work fine, except that you lose some enharmonic information. Sure -- my question was more along the lines of "is this...
> Is this resolved? Conceptually, yes. It would still be good to have a PR to improve the relevant documentation for this issue though.
mir_eval doesn't do this directly, but I have implemented this kind of thing over in my data augmentation library [muda](https://github.com/bmcfee/muda/), specifically here: https://github.com/bmcfee/muda/blob/9e4ef5fabce1f35e53829940aff51aa6b0a30893/muda/deformers/pitch.py#L18-L50 This operates on chord annotations that would...
I don't think I understand the suggested modification. Can you provide an example input and output of the proposed behavior?
This is because we use raw column headers from the csv input as dom element ids and css classes. It of course isn't helped by our conversion of numerical values...
This issue may be obviated by the discussion in #12. If we rewrite the interface to avoid exposing/requiring binarized columns, much of this will become simpler. The real resolution of...
Please include the output of `numpy.show_config()`. I suspect this is going to boil down to discrepancies in the underlying BLAS implementation.
Yes, it could be replaced. Is there a reason to do that though?