Keywords

large-scale audio corpora, phonetic analysis of formant measurements, DARLA and FAVE processing, challenges in corpus building, time and effort in transcription

Abstract

Large-scale transcribed audio corpora are available on Buckeye Corpus, Santa Barbara Corpus, etc.

How do these come to be? What’s the on-the-ground process of building such a corpus?

Here we discuss:

Methods for large-scale transcription
Early data & analysis resulting from transcription

Large-scale transcription:

Time to transcribe. Estimated: 10:1; Reality:13:1

Phonetic Analysis:

Comparison of formant measurements
In-house Praat script no good
DARLA filtered out 53%
Too early to tell if FAVE modifications were better

Original Publication Citation

Rachel Olsen, Michael Olsen, Joseph A. Stanley & Margaret E. L. Renwick. “Transcribing the Digital Archive of Southern Speech: Methods and Preliminary Analysis.” 84th Meeting of the Southeastern Conference on Linguistics (SECOL84). Charleston, SC. March 8–11, 2017.