Pareto Frontier Code

A Pareto frontier, also known as Pareto efficient or Pareto optimal frontier, is a concept in economics and decision theory that represents the optimal allocation of resources in a given system. It shows the combinations of two or more parameters that provide the best possible outcome, where no further improvement can be achieved for one parameter without worsening the other parameters. 

In our context, the first metric (𝑀1) is the difference among results from the ASR systems under test. This metric is used for measuring how likely a test case can expose failures. The second metric (𝑀2) is to directly measure the difference between a test case and the original audio. The rationale is to reduce the chance for the content of the test case to become malformed.

See wiki for more details.