Anti aliasing training wav, using various level sine sweeps 20Hz - 24 KHz added to the training wav

[Percepton]

Thanks for taking the time to write a feature request! Use the following prompts to help you describe what you're looking for. The more info you provide, the easier it'll be for me to understand what I can do to make it happen for you without having to come back and ask you questions.

Is your feature request related to a problem? Please describe.
A clear and concise description of what the problem is. Ex. I'm always frustrated when [...]

Describe the solution you'd like
A clear and concise description of what you want to happen.

Describe alternatives you've considered
A clear and concise description of any alternative solutions or features you've considered.

Additional context
Add any other context or screenshots about the feature request here.

[sdatkinson]

The current input.wav has exactly these sorts of sine sweeps at several levels from 0:12 to 0:15.

I'm aware of some work that's been done with various so-called "super inputs". It's somewhat interesting to me, but I need to think through what an acceptable evaluation might look like. As a simple example, increasing the length of the standardized training files is undesirable.

As of now, this is a lower priority because:

1. I'm not convinced that there's a practical benefit: v3_0_0.wav provides users with results that place NAM as the state of the art, and I don't see real evidence that the majority of users are dissatisfied even with inferior modeling tech (i.e. all of the alternatives).
2. This repo already provides the tools to use different models, data, and learning setups via nam-full; anyone interested in integrating this into the standardized trainers can easily do so by making an (see [FEATURE] Extensions #440).

In the meantime, if you have suggestions on a proposed evaluation, I wouldn't mind if you updated the OP to provide more details on the problem and how to determine that it's solved. [NB: I'm aware that playing sine sweeps through the plugin can elicit pseudo-aliasing artifacts. That isn't by itself very interesting to me because that doesn't reflect a realistic use case--that's not what a guitar sounds like, and I'm not surprised that a neural network doesn't do what you might expect when given irrelevant inputs. I'm looking for improvements that musicians and their listeners will appreciate while playing/listening to music with NAM.]