Added SpeechCommands example

[kaiidams]

This model achieves around 60-70% accuracy which is similar to PyTorch version. However, it depends on torchaudio.transform.Resample() which is not available in TorchSharp yet.

[NiklasGustafsson]

@kaiidams -- I merged the Resample PR, so if that means that this is no longer "WIP," then let me know and I'll do a review.

[NiklasGustafsson]

@kaiidams -- what are the system requirements for running the example on a GPU? Does it fit in 8GB of memory? That's been our general guideline for examples, to be able to run in a consumer-grade GPU.

[NiklasGustafsson]

@kaiidams -- sorry, I wasn't sure if you wanted to remove the "WIP" designation, but it seemed like you had done the change that you were blocked on. Please let me know if you want me to merge.

[kaiidams]

@kaiidams -- what are the system requirements for running the example on a GPU? Does it fit in 8GB of memory? That's been our general guideline for examples, to be able to run in a consumer-grade GPU.

@NiklasGustafsson This runs with CPU with 8GB RAM, though I haven't tested with GPU.

[NiklasGustafsson]

@kaiidams -- so, should I merge, or is this still work in progress?

UPDATE: never mind, I saw you took the "WIP" off the title. I'll merge