Table 2

Description of nine speech databases and their subsidiaries

Rank	Corpus	Year	Language	Average age	Subjects, n	Speech length (hour)	Location	Media type	Description	Citations, n
1	AphasiaBank					12				170
	Cantonese	2016	Cantonese	–	7/2		Hong Kong, China	Video	Native Cantonese speakers with stroke-induced aphasia
	Croatian	2017	Croatian	–	10/10		Zagreb, Croatia	Video	Native Croatian speakers with stroke-induced aphasia
	French	2016	French	–	11/14		France	Audio	Native French speakers with aphasia
	Italian	2011	Italian	–	10		USA	Video	Native Italian speakers with aphasia
	Mandarin	2015	Mandarin	45	9		China	Video	All patients with Mandarin as L1 and the aetiology is cerebral vascular accident (CVA)
	Spanish	2011	Spanish	–	4		USA	Video	Communication impairments by monolingual and bilingual speakers of Spanish and/or English
2	WRAP		English	54	64/200	30–40	USA	Audio	Connected speech problems of patients with dementia	168
3	Orozco-Arroyave Database	2014	Spanish	62; 60	50/50	>150	Spain	Audio	Speech recordings of patients with Parkinson’s disease and healthy controls	57
4	DementiaBank					70–80				17
	English Holland	2016	English	68; 72	2		USA	Video	Individuals with Alzheimer's disease—language tasks from a Telerounds presentation
	English Kempler	2016	English	81	6		USA	Audio	Individuals with Alzheimer's disease—conversation and Cookie-Theft picture descriptions
	English Pitt	2016	English	–	208/104		USA	Audio	Dementia and control data for four language tasks from a large longitudinal study
	English PPA DePaul	2016	English	66	1		USA	Video	Individual with primary progressive aphasia longitudinal data
	English PPA Hopkins	2016	English	–	36		USA	Audio	Individuals with primary progressive aphasia data
	German PPA	2016	English	–	–		Germany	Audio	Primary progressive aphasia data
	Mandarin_Lu	2016	Mandarin	–	52		China	Audio	Individuals with dementia data
	Spanish PerLA	2012	Spanish	–	21		Spain	Audio	Individuals with Alzheimer's disease and dementia data
	Taiwanese_Lu	2016	Taiwanese	–	16		China	Audio	Individuals with dementia
5	Cambridge Cookie-Theft Corpus	2010	English	54	87/227	41.5	Cambridge	Audio, brain scans	Individuals who have suffered from brain injury given the language task of picture description	9
6	CoDAS	2006	Dutch	54	6	0.5	Netherlands and Flanders	Audio	A pilot study of six aphasic speakers with two levels of annotation: an orthographic-phonetic transcription and a part-of-speech (POS) tagging	2
7	GREECAD	2016	Greek	55	72/28	1	Athens	Audio	An annotated Greek Corpus of Aphasic Discourse	1
8	FluencyBank					46				1
	POLER	2013	English	7	25/25		Washington, DC	Audio	Children with epilepsy and controls
	IISRP	2005	English	4	100/50		USA	Audio	Seminal study of children who stutter with controls
	Ratner	2012	English	3	23/15		USA	Audio	Children who stutter and controls
	Voices	2006	English	42	12		USA	Video	Interviews from the Voices of Stuttering project
	Ulm	1997	German	6	94		Ulm	Audio	Children who stutter from Ulm
9	DAIC-WOZ	2014	English	–	–	50	USA	Video	Anxiety, depression and post-traumatic stress disorder in University of Southern California	124

Year denotes the establishment of the speech database.
n refers to the number of subjects and citations.
– denotes information not available or the project is still ongoing with an increasing number of subjects.
/ denotes the number of patients proportioned to the number of controls.
CoDAS, Corpus of Dutch Aphasic Speech; DAIC-WOZ, Distress Analysis Interview Corpus-Wizard of Oz; GREECAD, Greek Corpus of Aphasic Discourse; POLER, Plasticity of Language in Epilepsy Research; WRAP, Wisconsin Registry for Alzheimer’s Prevention.