Introduces an exponential fit for speech mode. This means the scores will differ when using speech mode, but not audio mode. As a result, the conformance version was bumped to 310, and we will apply a v3.1.0 tag.
Merges internal Status{,Or} changes from @gjasny's PR #23. In many cases I preferred the internal version even though #23 had better formatting, because it will be overwritten by the next internal export due to things like alphabetical ordering or different internal namings.