Improve documentation for the resample function

Environment

0.10.0.dev0 https://librosa.org/doc/main/generated/librosa.resample.html

Describe the bug

There is an inconsistency among the res_type options for the resample function. While some are safe to use without an anti-aliasing filter, others are not.

The ones that are not safe include:

• soxr_qq
• polyphase
• linear
• zero_order_hold

To Reproduce

Run the following notebook. https://colab.research.google.com/drive/1VNS8O494CYocHrilUc7lZ_hKltNqb_PG?usp=sharing

Suggested fix

I suggest improving the documentation with a more explicit mention of the options that require anti-aliasing filtering. Not just stating that the quality is low. I will create a PR for this.

Sidenote: The documentation text talks about the kaiser_best being the default option, while the code states that soxr_hq is the default. This should probably be fixed in the same go.

[jpaasen]

Yes. I’m on it

[jpaasen]

Do you mean to ask if soxr_hq is anti-aliasing? The answer is yes. (We should also drop the “relatively slow” language, since soxr_hq is quite speedy.)

Sorry. Yes, I was thinking about the “relatively slow” wording.