JuliaMusic / MusicManipulations.jl

Manipulate music data, humanize, quantize and analyze music performances with Julia
MIT License
46 stars 7 forks source link

Humanizer via addition of pink-noise #56

Closed Datseris closed 4 years ago

Datseris commented 4 years ago

Research shows that human timing fluctuations are not white noise, but instead pink noise with varying exponents from -0.5 to -1.5.

A simple function humanize! could first generate such noise using e.g. ARFIMA.jl and then add it to the note positions.

michiboo commented 4 years ago

Hello! can I have a go at this issue?

michiboo commented 4 years ago

@Datseris I have created a PR for this issue, but I am not sure if I got it correctly, please review it when you have time! Thanks!

Datseris commented 4 years ago

Hello! can I have a go at this issue?

Hi, absolutely! Why not?!

michiboo commented 4 years ago

@Datseris I don't get how I can get the varying exponents from -0.5 to -1.5 from arfima.jl... do I pass it as parameters?

Datseris commented 4 years ago

Nope. I don't know either :D ARFIMA with fractional value for d can generate power-law correlated timeseries, but to my knolwedge, there isn't a "simple" way to connect the ARFIMA parameters with the power-law slope. Also, ARFIMA doesn't really generate "pure" power-laws...

michiboo commented 4 years ago

ah ok I will try to see what I can get! :)

Datseris commented 4 years ago

Don't worry too much about getting a perfect power-law for now. Let's first get the interface going well, and then we can think about the specific noise in detail.

michiboo commented 4 years ago

@Datseris Hi is the ARFIMA package on JuliaRegistries? If not I could try open a PR to add JuliaRegistries, at the moment there is some CI issues with the package as it cannot found it on the Registries.

Datseris commented 4 years ago

Hi @michiboo ,

Nope, it is not registered yet. Sorry for forgetting this! I'll register today (but it will take a couple of days until its public). In the meantime you can address the rest of the comments in #57 and I'll get to it.

Also, do you mind if we continue this discuss in PR #57 ? It makes it easier for me to stay organized :)

michiboo commented 4 years ago

@Datseris Ok thanks! I will discuss on there.