As of Oct 29, 2018 approx 2100-2200hrs US-EST, the IBM Watson US https://stream.watsonplatform.net/speech-to-text/api/v1/recognitions endpoint requires POST big audio file uploads with FileEntity to have setChunked(true) hint set on the FileEntity.
This is related to IBM Watson speech-to-text endpoint Oct 29, 2018 9pm US EST upgrade, that also included deprecation of older TLS protocals.
Old Opencast 1x version require additional upgrade on the IBM Watson transcription module. Newer Opencast 4x+ support the later TLS but require the setChucked(true) hint attribute set on the POSTed FileEntity.
The error manifests in the logs as:
"TranscriptionServiceException: Could not create recognition job. Status returned: 413"