DARLA:
Dartmouth Linguistic Automation
A suite of vowel formant
extraction tools tailored to research questions in
sociophonetics.
Semi-Automated Alignment and Extraction
Audio + Transcriptions → Formants
Similar to the FAVE pipeline (but using the Montreal Forced Aligner instead of FAVE-Align) with additional input and output options. Designed for research goals that require very accurate transcriptions.
Completely Automated Vowel Extraction
Audio → Transcriptions, Formants
This automated method is designed for "big data" research projects where a general vowel space is prioritized over word-level transcription accuracy.
NEW! We are happy to present Bed Word, a new transcription service that uses industry-modern models for the best possible transcription quality.
Users have the option to choose between DARLA's free in-house ASR model, or set up a paid Deepgram account for Bed Word. (DARLA is a completely free service, but Bed Word uses
a third-party company called Deepgram that has its own fee. We do not make any money from this tool.)
About
DARLA is designed and implemented by Sravana Reddy and James Stanford at Dartmouth. Bed Word is designed by Marcus Ma under the advisory of Lelia Glass, Linguistics Coordinator at Georgia Tech.
Our system makes use of automatic speech recognition using the CMU Sphinx, FAVE-Extract, Montreal Forced Aligner, and the R Vowels package to automate vowel analysis for sociophonetics.