InterSpeech 2021

CLAC: A Speech Corpus Of Healthy English Speakers
(3 minutes introduction)

R’mani Haulcy (MIT, USA), James Glass (MIT, USA)
This paper introduces the Crowdsourced Language Assessment Corpus (CLAC), a speech corpus consisting of audio recordings and automatically-generated transcripts for several speech and language tasks, as well as metadata for each of the speakers. The CLAC was created to provide the community with a collection of audio samples from various speakers that could be used to learn a general representation for speech from healthy subjects, as well as complement other health-related speech datasets, which tend to be limited. In this paper, we describe the data collection protocol and summarize the contents of the dataset. We also extract timing metrics from the recordings of each task to explore what those metrics look like for a large, English-speaking population. Lastly, we provide an example of how the dataset can be used by comparing the metrics to those extracted from a small sample of Frontotemporal Dementia subjects. We hope that this dataset will help advance the state of the art in the health and speech domain.