Which one fits your show?

The benchmark measured twenty-one systems on one hour of one podcast. That does not make any of them the best — it makes them better or worse at different things. Give us your feed and two answers, and we will say which of them suits the show you actually make, using those figures and nothing else.

Every number here comes from the benchmark itself — one hour of one podcast, measured against a transcript settled by ear, with the reference, the outputs and the scoring code published alongside it.