Here's the documentation for CNVkit's formats: http://cnvkit.readthedocs.io/en/stable/fileformats.html#target-and-antitarget-bin-level-coverages-cnn

In a nutshell, these are just tabular formats with column headers that can be easily loaded in R or Python (pandas).

• reference.cnn is the copy number control used for normalization of test samples.
• .cnr means normalized copy ratios. These are processed test samples, already normalized to the reference.
• .cns means copy number segments. These are the result of segmenting .cnr files, essentially where bins with similar estimated copy number are joined together.