Discrete Buffers for Multiple Buf Analyses

Getting Started and Wayfinding
1

Hi there -

I was hoping to clarify something about the example used in tedmoore’s Supercollider Corpus manipulation tutorials.

In the 3rd plotter example, we look at the use of multiple FluidBuf analyses methods as each one is send to a new buffer for statistic recording (FluidBufStat) and then another buffer (FluidBufCompose) to store information on successive frames.

In the 5th plotter example, we get 13 channels from FluidBufMFCC, we use a single FluidBufStats to analyze them, and then we pass the multichannel buffer to FluidBufFlatten, to get things knocked down to a single analyses point.

I’m curious about instances using multiple discrete FluidBuf analyses, as covered in Plotter 3 and 5 examples. .

Is the best practice to create individually-named buffers for each analysis, each statistic, and each point? Or does it make more sense to consolidate some of these into a single, multiple channel buffer at some point?

Thanks!

2

Now, this is really about what you prefer. I like many buffers, as they make it clear what is what. You can query them too, to see if they have the shape (nframes x nchans ) you expect.

What is not about preference, is what you need to send to addPoint - it requires a buffer of 1 and only 1 channel. A long flat thing. So how you get there is up to you, but each frame is then a dimension in your DataSet.

I hope this helps.

1 Like