Speech-to-text and ASR systems compared

A disclosed local transcription experiment across approved clips and a fixed owner recording. Compare corrected-reference word error rate, execution mode, hardware, and limitations without mixing these results with vendor claims or unrelated test sets.