Joll59 / d-ser-t

d-ser-t quantifies speech recognition accuracy of the MSFT speech service and/or user created MSFT custom speech service models.
2 stars 2 forks source link

only use false negatives when it exists in expectedUtterance #64

Closed ovishesh closed 5 years ago

ovishesh commented 5 years ago

Right now the logic checks for "keys" in each actualTranscription, as opposed to checking for the "value" in each expected, or at least both. I strongly feel that we should only be doing a replace if the expectedTranscription has that value in it.

HobbyProjects commented 5 years ago

Please consider adding a test for the new scenario. In fact, your change will break existing tests.