Skip to main content
Core-shed dictation gets a decoder that checks the depths

Core-shed dictation gets a decoder that checks the depths

Associate Professor Ragnhild Uzoma and Dr Vigdis Emenike bring speech recognition into the core shed, where a logger's spoken numbers meet the depths a borehole allows.

A paper from the School of Underlying Conditions holds speech recognition to the arithmetic of a drill-core run, and flags the runs it cannot make add up

The School of Underlying Conditions has made a habit of asking what a record is entitled to say about the ground, and its newest paper puts that question to the voice recorder. The Depths Must Add Up looks at the drill-core log as it is actually made, spoken aloud by a geologist with both hands in a tray, and at what happens to its numbers on the way to the page.

Writing this paper taught the School a good deal about how it talks when both hands are full. We are glad to have listened to ourselves so closely.

— Associate Professor Ragnhild Uzoma, Lead of the Basement Relations Group

Associate Professor Uzoma wrote the paper with Dr Vigdis Emenike, Senior Lecturer in the School. Their approach aligns recognition technology, field practice and the everyday arithmetic of drilling within a single framework for dictated logs, so that what a recogniser hears is weighed against what the hole itself permits, and the moments where the two part company are surfaced rather than smoothed over.

“A record that stops to ask a question is a record in good health,” said Dr Emenike. “There will be more of those questions in the core shed, and the School is glad to be the one hearing them.”

The paper is the School’s first contribution to speech and audio research, and the University counts it a natural extension of an agenda that has always begun by paying attention.

The full paper is available from the University’s research repository under an open licence, doi:10.5555/slop.5ib0r5.