Colin Raffel

Nice! I am very excited for this. I have https://github.com/craffel/lstm_benchmarks, which is an out-of-date benchmark of Theano vs. rnnlib vs. currennt (which, at the time I wrote the benchmarks, were...

> (especially if they're implemented in Lasagne). That would be nice :) though I am personally interested in which of the Theano-based libraries manage to eek out the most performance,...

> @craffel, couldn't get what do you meant by this This was in response to > One thing I've also been thinking about like @daviddao is how to validate that...

Yeah, maybe librosa.fb or something. I could see that making sense although I think most librosa users will only use fb's when they are computing features... who knows though.

I don't think the different filterbank generators need to have the same parameters. Are you feeling like it needs to be a single function? Why can't they be different functions?

Oh, I was thinking like I was discussing above - you have separate functions which compute filterbanks with different shapes, with arbitrary center frequencies. Then, you use those filterbanks to...

I don't think we're talking about the same thing - to me computing a chroma filterbank is equivalent to the basic filter shape functions, just with a different spacing from,...

I like this idea, but don't have any input on the best method.

Thanks.

I have the code from the LSTM benchmark, which needs a little updating, which is a noisy speech recognition experiment. It would be a little silly to host it in...